{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "## 필요한 모듈 import" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import pandas as pd\n", "import numpy as np\n", "from sklearn.pipeline import FeatureUnion\n", "from imblearn.pipeline import make_pipeline, Pipeline\n", "from sklearn.preprocessing import MinMaxScaler\n", "from sklearn.model_selection import StratifiedKFold, GridSearchCV, RandomizedSearchCV, train_test_split\n", "from sklearn.decomposition import PCA\n", "from sklearn.discriminant_analysis import LinearDiscriminantAnalysis" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "from sklearn.feature_selection import VarianceThreshold" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "APCFP = pd.read_csv('finalfingerprint_sub_apc.csv')" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
target_name_bintarget_name_tertarget_name_quatersmilesAPC2D1_C_CAPC2D1_C_NAPC2D1_C_OAPC2D1_C_SAPC2D1_C_PAPC2D1_C_F...SubFPC303SubFPC304SubFPC305SubFPC306SubFPC307bond:C=S_carbonyl_thio_genericbond:CN_amine_pri-NH2_aromaticbond:CS_sulfidebond:S=O_sulfonyl_genericring:hetero_[6_6]_O_benzopyrone_(1_4-)
0AAA1Nc1ccccc1610000...0000601000
1AAA1COc1ccc(N)cc1612000...0000601000
2AAA1COP(=S)(OC)SCn1nnc2ccccc2c1=O733100...0000900100
3AAA1CC(C)(c1ccc(O)cc1)c1ccc(O)cc11602000...00001300000
4AAA1CC(C)(C)c1cc(O)ccc1O1002000...0000700000
..................................................................
582BCCCSCCCCN=C=S320300...0000010100
583BCCCCCN=C=S220100...0000010000
584BCCCS(=O)CCCCC#N410200...0000100000
585BCCCSCCCCC#N410200...0000000100
586BCCC=CCSC#N210200...0000100100
\n", "

587 rows × 1096 columns

\n", "
" ], "text/plain": [ " target_name_bin target_name_ter target_name_quater \\\n", "0 A A A1 \n", "1 A A A1 \n", "2 A A A1 \n", "3 A A A1 \n", "4 A A A1 \n", ".. ... ... ... \n", "582 B C C \n", "583 B C C \n", "584 B C C \n", "585 B C C \n", "586 B C C \n", "\n", " smiles APC2D1_C_C APC2D1_C_N APC2D1_C_O \\\n", "0 Nc1ccccc1 6 1 0 \n", "1 COc1ccc(N)cc1 6 1 2 \n", "2 COP(=S)(OC)SCn1nnc2ccccc2c1=O 7 3 3 \n", "3 CC(C)(c1ccc(O)cc1)c1ccc(O)cc1 16 0 2 \n", "4 CC(C)(C)c1cc(O)ccc1O 10 0 2 \n", ".. ... ... ... ... \n", "582 CSCCCCN=C=S 3 2 0 \n", "583 CCCN=C=S 2 2 0 \n", "584 CS(=O)CCCCC#N 4 1 0 \n", "585 CSCCCCC#N 4 1 0 \n", "586 C=CCSC#N 2 1 0 \n", "\n", " APC2D1_C_S APC2D1_C_P APC2D1_C_F ... SubFPC303 SubFPC304 SubFPC305 \\\n", "0 0 0 0 ... 0 0 0 \n", "1 0 0 0 ... 0 0 0 \n", "2 1 0 0 ... 0 0 0 \n", "3 0 0 0 ... 0 0 0 \n", "4 0 0 0 ... 0 0 0 \n", ".. ... ... ... ... ... ... ... \n", "582 3 0 0 ... 0 0 0 \n", "583 1 0 0 ... 0 0 0 \n", "584 2 0 0 ... 0 0 0 \n", "585 2 0 0 ... 0 0 0 \n", "586 2 0 0 ... 0 0 0 \n", "\n", " SubFPC306 SubFPC307 bond:C=S_carbonyl_thio_generic \\\n", "0 0 6 0 \n", "1 0 6 0 \n", "2 0 9 0 \n", "3 0 13 0 \n", "4 0 7 0 \n", ".. ... ... ... \n", "582 0 0 1 \n", "583 0 0 1 \n", "584 0 1 0 \n", "585 0 0 0 \n", "586 0 1 0 \n", "\n", " bond:CN_amine_pri-NH2_aromatic bond:CS_sulfide \\\n", "0 1 0 \n", "1 1 0 \n", "2 0 1 \n", "3 0 0 \n", "4 0 0 \n", ".. ... ... \n", "582 0 1 \n", "583 0 0 \n", "584 0 0 \n", "585 0 1 \n", "586 0 1 \n", "\n", " bond:S=O_sulfonyl_generic ring:hetero_[6_6]_O_benzopyrone_(1_4-) \n", "0 0 0 \n", "1 0 0 \n", "2 0 0 \n", "3 0 0 \n", "4 0 0 \n", ".. ... ... \n", "582 0 0 \n", "583 0 0 \n", "584 0 0 \n", "585 0 0 \n", "586 0 0 \n", "\n", "[587 rows x 1096 columns]" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "APCFP" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "target_bin = APCFP.loc[:, 'target_name_bin']" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "target_bin = pd.DataFrame(target_bin)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [], "source": [ "target_ter = APCFP.loc[:, 'target_name_ter']" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "target_ter = pd.DataFrame(target_ter)" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [], "source": [ "target_quater = APCFP.loc[:, 'target_name_quater']\n", "target_quater = pd.DataFrame(target_quater)" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
target_name_ter
0A
1A
2A
3A
4A
......
582C
583C
584C
585C
586C
\n", "

587 rows × 1 columns

\n", "
" ], "text/plain": [ " target_name_ter\n", "0 A\n", "1 A\n", "2 A\n", "3 A\n", "4 A\n", ".. ...\n", "582 C\n", "583 C\n", "584 C\n", "585 C\n", "586 C\n", "\n", "[587 rows x 1 columns]" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "target_ter" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
target_name_quater
0A1
1A1
2A1
3A1
4A1
......
582C
583C
584C
585C
586C
\n", "

587 rows × 1 columns

\n", "
" ], "text/plain": [ " target_name_quater\n", "0 A1\n", "1 A1\n", "2 A1\n", "3 A1\n", "4 A1\n", ".. ...\n", "582 C\n", "583 C\n", "584 C\n", "585 C\n", "586 C\n", "\n", "[587 rows x 1 columns]" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "target_quater" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "features = APCFP.loc[:, 'APC2D1_C_C':'ring:hetero_[6_6]_O_benzopyrone_(1_4-)']" ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "scrolled": true }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
APC2D1_C_CAPC2D1_C_NAPC2D1_C_OAPC2D1_C_SAPC2D1_C_PAPC2D1_C_FAPC2D1_C_ClAPC2D1_C_BrAPC2D1_C_IAPC2D1_C_B...SubFPC303SubFPC304SubFPC305SubFPC306SubFPC307bond:C=S_carbonyl_thio_genericbond:CN_amine_pri-NH2_aromaticbond:CS_sulfidebond:S=O_sulfonyl_genericring:hetero_[6_6]_O_benzopyrone_(1_4-)
06100000000...0000601000
16120000000...0000601000
27331000000...0000900100
316020000000...00001300000
410020000000...0000700000
..................................................................
5823203000000...0000010100
5832201000000...0000010000
5844102000000...0000100000
5854102000000...0000000100
5862102000000...0000100100
\n", "

587 rows × 1092 columns

\n", "
" ], "text/plain": [ " APC2D1_C_C APC2D1_C_N APC2D1_C_O APC2D1_C_S APC2D1_C_P APC2D1_C_F \\\n", "0 6 1 0 0 0 0 \n", "1 6 1 2 0 0 0 \n", "2 7 3 3 1 0 0 \n", "3 16 0 2 0 0 0 \n", "4 10 0 2 0 0 0 \n", ".. ... ... ... ... ... ... \n", "582 3 2 0 3 0 0 \n", "583 2 2 0 1 0 0 \n", "584 4 1 0 2 0 0 \n", "585 4 1 0 2 0 0 \n", "586 2 1 0 2 0 0 \n", "\n", " APC2D1_C_Cl APC2D1_C_Br APC2D1_C_I APC2D1_C_B ... SubFPC303 \\\n", "0 0 0 0 0 ... 0 \n", "1 0 0 0 0 ... 0 \n", "2 0 0 0 0 ... 0 \n", "3 0 0 0 0 ... 0 \n", "4 0 0 0 0 ... 0 \n", ".. ... ... ... ... ... ... \n", "582 0 0 0 0 ... 0 \n", "583 0 0 0 0 ... 0 \n", "584 0 0 0 0 ... 0 \n", "585 0 0 0 0 ... 0 \n", "586 0 0 0 0 ... 0 \n", "\n", " SubFPC304 SubFPC305 SubFPC306 SubFPC307 \\\n", "0 0 0 0 6 \n", "1 0 0 0 6 \n", "2 0 0 0 9 \n", "3 0 0 0 13 \n", "4 0 0 0 7 \n", ".. ... ... ... ... \n", "582 0 0 0 0 \n", "583 0 0 0 0 \n", "584 0 0 0 1 \n", "585 0 0 0 0 \n", "586 0 0 0 1 \n", "\n", " bond:C=S_carbonyl_thio_generic bond:CN_amine_pri-NH2_aromatic \\\n", "0 0 1 \n", "1 0 1 \n", "2 0 0 \n", "3 0 0 \n", "4 0 0 \n", ".. ... ... \n", "582 1 0 \n", "583 1 0 \n", "584 0 0 \n", "585 0 0 \n", "586 0 0 \n", "\n", " bond:CS_sulfide bond:S=O_sulfonyl_generic \\\n", "0 0 0 \n", "1 0 0 \n", "2 1 0 \n", "3 0 0 \n", "4 0 0 \n", ".. ... ... \n", "582 1 0 \n", "583 0 0 \n", "584 0 0 \n", "585 1 0 \n", "586 1 0 \n", "\n", " ring:hetero_[6_6]_O_benzopyrone_(1_4-) \n", "0 0 \n", "1 0 \n", "2 0 \n", "3 0 \n", "4 0 \n", ".. ... \n", "582 0 \n", "583 0 \n", "584 0 \n", "585 0 \n", "586 0 \n", "\n", "[587 rows x 1092 columns]" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "features" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [], "source": [ "from imblearn.over_sampling import *\n", "from sklearn.neural_network import MLPClassifier" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# AtomPairCount_Binary" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [], "source": [ "seed=28\n", "X_train_bin, X_test_bin, y_train_bin, y_test_bin = train_test_split(features, target_bin, test_size=0.2, random_state=seed, stratify=target_bin)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "PCA" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [], "source": [ "pipe_ann_pca = Pipeline([(\"std\", MinMaxScaler()),\n", " (\"thresholder\", VarianceThreshold(threshold=0.01)),\n", " (\"PCA\", PCA()),\n", " (\"smote\", SMOTE(random_state=seed)),\n", " (\"classifier\", MLPClassifier(random_state=seed, max_iter=10000))])" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [], "source": [ "random_grid_pca = [{\"PCA__n_components\": [1, 2, 3],\n", " \"classifier__hidden_layer_sizes\": [(10,), (50,), (100,), (200,)],\n", " \"classifier__activation\": ['relu', 'logistic', 'tanh', 'identity'],\n", " \"classifier__solver\": ['lbfgs', 'adam', 'sgd'],\n", " \"classifier__alpha\": [0.00001, 0.001, 0.1, 10]}]" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [], "source": [ "cv = StratifiedKFold(n_splits=5, random_state=seed, shuffle=True)" ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [], "source": [ "rand_ann_pca = RandomizedSearchCV(pipe_ann_pca, random_grid_pca, n_iter=100, cv=cv, random_state=seed, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 13.8s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 46.0s\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 1.9min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 4.0min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True,\n", " iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=Fal...\n", " param_distributions=[{'PCA__n_components': [1, 2, 3],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.fit(X_train_bin, y_train_bin.values.ravel())" ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.5521919013688538" ] }, "execution_count": 21, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.best_score_" ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'lbfgs',\n", " 'classifier__hidden_layer_sizes': (100,),\n", " 'classifier__alpha': 1e-05,\n", " 'classifier__activation': 'logistic',\n", " 'PCA__n_components': 3}" ] }, "execution_count": 22, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.best_params_" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [], "source": [ "param_grid_pca_bin = [{\"PCA__n_components\": [3],\n", " \"classifier__hidden_layer_sizes\": [(100,), (100, 100,), (100, 100, 100,), (100, 100, 100, 100,), (100, 100, 100, 100, 100,)],\n", " \"classifier__activation\": ['logistic'],\n", " \"classifier__solver\": ['lbfgs'],\n", " \"classifier__alpha\": [0.00001, 0.00005, 0.0001, 0.0005]}]" ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [], "source": [ "clf_ann_pca_bin = GridSearchCV(pipe_ann_pca, param_grid_pca_bin, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 20 candidates, totalling 100 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 50.5s\n", "[Parallel(n_jobs=-1)]: Done 100 out of 100 | elapsed: 3.7min finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True, iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=False)),\n", " (...\n", " param_grid=[{'PCA__n_components': [3],\n", " 'classifier__activation': ['logistic'],\n", " 'classifier__alpha': [1e-05, 5e-05, 0.0001, 0.0005],\n", " 'classifier__hidden_layer_sizes': [(100,), (100, 100),\n", " (100, 100, 100),\n", " (100, 100, 100,\n", " 100),\n", " (100, 100, 100,\n", " 100, 100)],\n", " 'classifier__solver': ['lbfgs']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 25, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_bin.fit(X_train_bin, y_train_bin.values.ravel())" ] }, { "cell_type": "code", "execution_count": 26, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'PCA__n_components': 3,\n", " 'classifier__activation': 'logistic',\n", " 'classifier__alpha': 1e-05,\n", " 'classifier__hidden_layer_sizes': (100, 100),\n", " 'classifier__solver': 'lbfgs'}" ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_bin.best_params_" ] }, { "cell_type": "code", "execution_count": 27, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.5789953331756281" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_bin.best_score_" ] }, { "cell_type": "code", "execution_count": 28, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4008\n" ] } ], "source": [ "from sklearn import metrics\n", "bin_pca_predict = clf_ann_pca_bin.predict(X_test_bin)\n", "testscore_bin_pca = metrics.f1_score(y_test_bin, bin_pca_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_bin_pca))" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[39 58]\n", " [ 9 12]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_bin, bin_pca_predict))" ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Binary_APC_ANN_PCA.pkl']" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import joblib\n", "joblib.dump(clf_ann_pca_bin, \"Binary_APC_ANN_PCA.pkl\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "LDA" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [], "source": [ "pipe_ann_lda = Pipeline([(\"std\", MinMaxScaler()),\n", " (\"thresholder\", VarianceThreshold(threshold=0.01)),\n", " (\"LDA\", LinearDiscriminantAnalysis()),\n", " (\"smote\", SMOTE(random_state=seed)),\n", " (\"classifier\", MLPClassifier(random_state=seed, max_iter=10000))])" ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [], "source": [ "random_grid_lda_bin = [{\"LDA__n_components\": [1],\n", " \"classifier__hidden_layer_sizes\": [(10,), (50,), (100,), (200,)],\n", " \"classifier__activation\": ['relu', 'logistic', 'tanh', 'identity'],\n", " \"classifier__solver\": ['lbfgs', 'adam', 'sgd'],\n", " \"classifier__alpha\": [0.00001, 0.001, 0.1, 10]}]" ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [], "source": [ "rand_ann_lda_bin = RandomizedSearchCV(pipe_ann_lda, random_grid_lda_bin, n_iter=100, cv=cv, random_state=seed, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 9.9s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 38.3s\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 56.9s\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 2.7min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,...\n", " param_distributions=[{'LDA__n_components': [1],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 23, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_bin.fit(X_train_bin, y_train_bin.values.ravel())" ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.6335683345630023" ] }, "execution_count": 24, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_bin.best_score_" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'lbfgs',\n", " 'classifier__hidden_layer_sizes': (200,),\n", " 'classifier__alpha': 1e-05,\n", " 'classifier__activation': 'logistic',\n", " 'LDA__n_components': 1}" ] }, "execution_count": 25, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_bin.best_params_" ] }, { "cell_type": "code", "execution_count": 26, "metadata": {}, "outputs": [], "source": [ "param_grid_lda_bin = [{\"LDA__n_components\": [1],\n", " \"classifier__hidden_layer_sizes\": [(200,), (200, 200,), (200, 200, 200,), (200, 200, 200, 200,), (200, 200, 200, 200, 200,)],\n", " \"classifier__activation\": ['logistic'],\n", " \"classifier__solver\": ['lbfgs'],\n", " \"classifier__alpha\": [0.00001, 0.00005, 0.0001, 0.0005, 0.001]}]" ] }, { "cell_type": "code", "execution_count": 27, "metadata": {}, "outputs": [], "source": [ "clf_ann_lda_bin = GridSearchCV(pipe_ann_lda, param_grid_lda_bin, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 28, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 25 candidates, totalling 125 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 22.9s\n", "[Parallel(n_jobs=-1)]: Done 125 out of 125 | elapsed: 41.9min finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,\n", " tol=0....\n", " param_grid=[{'LDA__n_components': [1],\n", " 'classifier__activation': ['logistic'],\n", " 'classifier__alpha': [1e-05, 5e-05, 0.0001, 0.0005,\n", " 0.001],\n", " 'classifier__hidden_layer_sizes': [(200,), (200, 200),\n", " (200, 200, 200),\n", " (200, 200, 200,\n", " 200),\n", " (200, 200, 200,\n", " 200, 200)],\n", " 'classifier__solver': ['lbfgs']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 28, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_bin.fit(X_train_bin, y_train_bin.values.ravel())" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'LDA__n_components': 1,\n", " 'classifier__activation': 'logistic',\n", " 'classifier__alpha': 1e-05,\n", " 'classifier__hidden_layer_sizes': (200,),\n", " 'classifier__solver': 'lbfgs'}" ] }, "execution_count": 29, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_bin.best_params_" ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.6335683345630023" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_bin.best_score_" ] }, { "cell_type": "code", "execution_count": 32, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.5851\n" ] } ], "source": [ "from sklearn import metrics\n", "bin_lda_predict = clf_ann_lda_bin.predict(X_test_bin)\n", "testscore_bin_lda = metrics.f1_score(y_test_bin, bin_lda_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_bin_lda))" ] }, { "cell_type": "code", "execution_count": 33, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[73 24]\n", " [11 10]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_bin, bin_lda_predict))" ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Binary_APC_ANN_LDA.pkl']" ] }, "execution_count": 35, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import joblib\n", "joblib.dump(clf_ann_lda_bin, \"Binary_APC_ANN_LDA.pkl\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Morgan_Ternary" ] }, { "cell_type": "code", "execution_count": 36, "metadata": {}, "outputs": [], "source": [ "X_train_ter, X_test_ter, y_train_ter, y_test_ter = train_test_split(features, target_ter, test_size=0.2, random_state=seed, stratify=target_ter)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "PCA" ] }, { "cell_type": "code", "execution_count": 37, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 41.7s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 2.1min\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 5.3min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 7.9min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True,\n", " iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=Fal...\n", " param_distributions=[{'PCA__n_components': [1, 2, 3],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 37, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.fit(X_train_ter, y_train_ter.values.ravel())" ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.4255029778502319" ] }, "execution_count": 38, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.best_score_" ] }, { "cell_type": "code", "execution_count": 39, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'lbfgs',\n", " 'classifier__hidden_layer_sizes': (10,),\n", " 'classifier__alpha': 1e-05,\n", " 'classifier__activation': 'tanh',\n", " 'PCA__n_components': 3}" ] }, "execution_count": 39, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.best_params_" ] }, { "cell_type": "code", "execution_count": 40, "metadata": {}, "outputs": [], "source": [ "param_grid_pca_ter = [{\"PCA__n_components\": [3],\n", " \"classifier__hidden_layer_sizes\": [(10,), (10, 10,), (10, 10, 10,), (10, 10, 10, 10,), (10, 10, 10, 10, 10,)],\n", " \"classifier__activation\": ['tanh'],\n", " \"classifier__solver\": ['lbfgs'],\n", " \"classifier__alpha\": [0.00005, 0.0001, 0.0005, 0.001, 0.005, 0.01, 0.05]}]" ] }, { "cell_type": "code", "execution_count": 41, "metadata": {}, "outputs": [], "source": [ "clf_ann_pca_ter = GridSearchCV(pipe_ann_pca, param_grid_pca_ter, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 42, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 35 candidates, totalling 175 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 46.1s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 4.6min\n", "[Parallel(n_jobs=-1)]: Done 175 out of 175 | elapsed: 5.6min finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True, iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=False)),\n", " (...\n", " iid='deprecated', n_jobs=-1,\n", " param_grid=[{'PCA__n_components': [3],\n", " 'classifier__activation': ['tanh'],\n", " 'classifier__alpha': [5e-05, 0.0001, 0.0005, 0.001,\n", " 0.005, 0.01, 0.05],\n", " 'classifier__hidden_layer_sizes': [(10,), (10, 10),\n", " (10, 10, 10),\n", " (10, 10, 10, 10),\n", " (10, 10, 10, 10,\n", " 10)],\n", " 'classifier__solver': ['lbfgs']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 42, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_ter.fit(X_train_ter, y_train_ter.values.ravel())" ] }, { "cell_type": "code", "execution_count": 43, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'PCA__n_components': 3,\n", " 'classifier__activation': 'tanh',\n", " 'classifier__alpha': 5e-05,\n", " 'classifier__hidden_layer_sizes': (10,),\n", " 'classifier__solver': 'lbfgs'}" ] }, "execution_count": 43, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_ter.best_params_" ] }, { "cell_type": "code", "execution_count": 44, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.453891322636644" ] }, "execution_count": 44, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_ter.best_score_" ] }, { "cell_type": "code", "execution_count": 45, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4030\n" ] } ], "source": [ "ter_pca_predict = clf_ann_pca_ter.predict(X_test_ter)\n", "testscore_ter_pca = metrics.f1_score(y_test_ter, ter_pca_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_ter_pca))" ] }, { "cell_type": "code", "execution_count": 46, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[34 25 12]\n", " [12 9 5]\n", " [ 5 7 9]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_ter, ter_pca_predict))" ] }, { "cell_type": "code", "execution_count": 47, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Ternary_APC_ANN_PCA.pkl']" ] }, "execution_count": 47, "metadata": {}, "output_type": "execute_result" } ], "source": [ "joblib.dump(clf_ann_pca_ter, 'Ternary_APC_ANN_PCA.pkl')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "LDA" ] }, { "cell_type": "code", "execution_count": 48, "metadata": {}, "outputs": [], "source": [ "random_grid_lda_ter = [{\"LDA__n_components\": [1, 2],\n", " \"classifier__hidden_layer_sizes\": [(10,), (50,), (100,), (200,)],\n", " \"classifier__activation\": ['relu', 'logistic', 'tanh', 'identity'],\n", " \"classifier__solver\": ['lbfgs', 'adam', 'sgd'],\n", " \"classifier__alpha\": [0.00001, 0.001, 0.1, 10]}]" ] }, { "cell_type": "code", "execution_count": 49, "metadata": {}, "outputs": [], "source": [ "rand_ann_lda_ter = RandomizedSearchCV(pipe_ann_lda, random_grid_lda_ter, n_iter=100, cv=cv, random_state=seed, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 50, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 15.6s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 1.3min\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 2.7min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 6.1min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,...\n", " param_distributions=[{'LDA__n_components': [1, 2],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 50, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_ter.fit(X_train_ter, y_train_ter.values.ravel())" ] }, { "cell_type": "code", "execution_count": 51, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.48399687241472683" ] }, "execution_count": 51, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_ter.best_score_" ] }, { "cell_type": "code", "execution_count": 52, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'lbfgs',\n", " 'classifier__hidden_layer_sizes': (200,),\n", " 'classifier__alpha': 0.1,\n", " 'classifier__activation': 'relu',\n", " 'LDA__n_components': 2}" ] }, "execution_count": 52, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_ter.best_params_" ] }, { "cell_type": "code", "execution_count": 53, "metadata": {}, "outputs": [], "source": [ "param_grid_lda_ter = [{\"LDA__n_components\": [2],\n", " \"classifier__hidden_layer_sizes\": [(200,), (200, 200,), (200, 200, 200,), (200, 200, 200, 200,), (200, 200, 200, 200, 200,)],\n", " \"classifier__activation\": ['relu'],\n", " \"classifier__solver\": ['lbfgs'],\n", " \"classifier__alpha\": [0.005, 0.01, 0.05, 0.1, 0.5, 1, 5]}]" ] }, { "cell_type": "code", "execution_count": 54, "metadata": {}, "outputs": [], "source": [ "clf_ann_lda_ter = GridSearchCV(pipe_ann_lda, param_grid_lda_ter, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 55, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 35 candidates, totalling 175 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 11.8min\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 146.4min\n", "[Parallel(n_jobs=-1)]: Done 175 out of 175 | elapsed: 169.5min finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,\n", " tol=0....\n", " iid='deprecated', n_jobs=-1,\n", " param_grid=[{'LDA__n_components': [2],\n", " 'classifier__activation': ['relu'],\n", " 'classifier__alpha': [0.005, 0.01, 0.05, 0.1, 0.5, 1,\n", " 5],\n", " 'classifier__hidden_layer_sizes': [(200,), (200, 200),\n", " (200, 200, 200),\n", " (200, 200, 200,\n", " 200),\n", " (200, 200, 200,\n", " 200, 200)],\n", " 'classifier__solver': ['lbfgs']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 55, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_ter.fit(X_train_ter, y_train_ter.values.ravel())" ] }, { "cell_type": "code", "execution_count": 56, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'LDA__n_components': 2,\n", " 'classifier__activation': 'relu',\n", " 'classifier__alpha': 0.1,\n", " 'classifier__hidden_layer_sizes': (200,),\n", " 'classifier__solver': 'lbfgs'}" ] }, "execution_count": 56, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_ter.best_params_" ] }, { "cell_type": "code", "execution_count": 57, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.48399687241472683" ] }, "execution_count": 57, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_ter.best_score_" ] }, { "cell_type": "code", "execution_count": 58, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4224\n" ] } ], "source": [ "from sklearn import metrics\n", "ter_lda_predict = clf_ann_lda_ter.predict(X_test_ter)\n", "testscore_ter_lda = metrics.f1_score(y_test_ter, ter_lda_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_ter_lda))" ] }, { "cell_type": "code", "execution_count": 59, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[44 17 10]\n", " [15 7 4]\n", " [ 3 10 8]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_ter, ter_lda_predict))" ] }, { "cell_type": "code", "execution_count": 60, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Ternary_APC_ANN_LDA.pkl']" ] }, "execution_count": 60, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import joblib\n", "joblib.dump(clf_ann_lda_ter, 'Ternary_APC_ANN_LDA.pkl')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Quaternary(w/o SMOTE)" ] }, { "cell_type": "code", "execution_count": 61, "metadata": {}, "outputs": [], "source": [ "X_train_quater, X_test_quater, y_train_quater, y_test_quater = train_test_split(features, target_quater, test_size=0.2, random_state=seed, stratify=target_quater)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "PCA" ] }, { "cell_type": "code", "execution_count": 62, "metadata": {}, "outputs": [], "source": [ "pipe_ann_pca_smotx = Pipeline([(\"std\", MinMaxScaler()),\n", " (\"thresholder\", VarianceThreshold(threshold=0.01)),\n", " (\"PCA\", PCA()),\n", " (\"classifier\", MLPClassifier(random_state=seed, max_iter=10000))])" ] }, { "cell_type": "code", "execution_count": 63, "metadata": {}, "outputs": [], "source": [ "rand_ann_pca_smotx = RandomizedSearchCV(pipe_ann_pca_smotx, random_grid_pca, n_iter=100, cv=cv, random_state=seed, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 64, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 35.3s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 1.4min\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 3.2min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 4.3min finished\n", "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\neural_network\\_multilayer_perceptron.py:470: ConvergenceWarning: lbfgs failed to converge (status=1):\n", "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n", "\n", "Increase the number of iterations (max_iter) or scale the data as shown in:\n", " https://scikit-learn.org/stable/modules/preprocessing.html\n", " self.n_iter_ = _check_optimize_result(\"lbfgs\", opt_res, self.max_iter)\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True,\n", " iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=Fal...\n", " param_distributions=[{'PCA__n_components': [1, 2, 3],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 64, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca_smotx.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 65, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.39999282896523203" ] }, "execution_count": 65, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca_smotx.best_score_" ] }, { "cell_type": "code", "execution_count": 66, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'lbfgs',\n", " 'classifier__hidden_layer_sizes': (50,),\n", " 'classifier__alpha': 0.001,\n", " 'classifier__activation': 'logistic',\n", " 'PCA__n_components': 3}" ] }, "execution_count": 66, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca_smotx.best_params_" ] }, { "cell_type": "code", "execution_count": 67, "metadata": {}, "outputs": [], "source": [ "param_grid_pca_quater = [{\"PCA__n_components\": [3],\n", " \"classifier__hidden_layer_sizes\": [(50,), (50, 50,), (50, 50, 50,), (50, 50, 50, 50,), (50, 50, 50, 50, 50,)],\n", " \"classifier__activation\": ['logistic'],\n", " \"classifier__solver\": ['lbfgs'],\n", " \"classifier__alpha\": [0.00005, 0.0001, 0.0005, 0.001, 0.005, 0.01, 0.05]}]" ] }, { "cell_type": "code", "execution_count": 68, "metadata": {}, "outputs": [], "source": [ "clf_ann_pca_smotx = GridSearchCV(pipe_ann_pca_smotx, param_grid_pca_quater, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 69, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 35 candidates, totalling 175 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 54.4s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 5.2min\n", "[Parallel(n_jobs=-1)]: Done 175 out of 175 | elapsed: 6.5min finished\n", "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\neural_network\\_multilayer_perceptron.py:470: ConvergenceWarning: lbfgs failed to converge (status=1):\n", "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n", "\n", "Increase the number of iterations (max_iter) or scale the data as shown in:\n", " https://scikit-learn.org/stable/modules/preprocessing.html\n", " self.n_iter_ = _check_optimize_result(\"lbfgs\", opt_res, self.max_iter)\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True, iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=False)),\n", " (...\n", " param_grid=[{'PCA__n_components': [3],\n", " 'classifier__activation': ['logistic'],\n", " 'classifier__alpha': [5e-05, 0.0001, 0.0005, 0.001,\n", " 0.005, 0.01, 0.05],\n", " 'classifier__hidden_layer_sizes': [(50,), (50, 50),\n", " (50, 50, 50),\n", " (50, 50, 50, 50),\n", " (50, 50, 50, 50,\n", " 50)],\n", " 'classifier__solver': ['lbfgs']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 69, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_smotx.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 70, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'PCA__n_components': 3,\n", " 'classifier__activation': 'logistic',\n", " 'classifier__alpha': 0.001,\n", " 'classifier__hidden_layer_sizes': (50,),\n", " 'classifier__solver': 'lbfgs'}" ] }, "execution_count": 70, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_smotx.best_params_" ] }, { "cell_type": "code", "execution_count": 71, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.39999282896523203" ] }, "execution_count": 71, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_smotx.best_score_" ] }, { "cell_type": "code", "execution_count": 72, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4372\n" ] } ], "source": [ "quater_pca_predict = clf_ann_pca_smotx.predict(X_test_quater)\n", "testscore_quater_pca = metrics.f1_score(y_test_quater, quater_pca_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_quater_pca))" ] }, { "cell_type": "code", "execution_count": 73, "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[14 8 2 1]\n", " [11 24 5 6]\n", " [ 2 9 9 6]\n", " [ 4 6 4 7]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_quater, quater_pca_predict))" ] }, { "cell_type": "code", "execution_count": 74, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Quaternary_APC_ANN_PCA_smotx.pkl']" ] }, "execution_count": 74, "metadata": {}, "output_type": "execute_result" } ], "source": [ "joblib.dump(clf_ann_pca_smotx, 'Quaternary_APC_ANN_PCA_smotx.pkl')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "LDA" ] }, { "cell_type": "code", "execution_count": 75, "metadata": {}, "outputs": [], "source": [ "random_grid_lda_quater = [{\"LDA__n_components\": [1, 2, 3],\n", " \"classifier__hidden_layer_sizes\": [(10,), (50,), (100,), (200,)],\n", " \"classifier__activation\": ['relu', 'logistic', 'tanh', 'identity'],\n", " \"classifier__solver\": ['lbfgs', 'adam', 'sgd'],\n", " \"classifier__alpha\": [0.00001, 0.001, 0.1, 10]}]" ] }, { "cell_type": "code", "execution_count": 76, "metadata": {}, "outputs": [], "source": [ "pipe_ann_lda_smotx = Pipeline([(\"std\", MinMaxScaler()),\n", " (\"thresholder\", VarianceThreshold(threshold=0.01)),\n", " (\"LDA\", LinearDiscriminantAnalysis()),\n", " (\"classifier\", MLPClassifier(random_state=seed, max_iter=10000))])" ] }, { "cell_type": "code", "execution_count": 77, "metadata": {}, "outputs": [], "source": [ "rand_ann_lda_smotx = RandomizedSearchCV(pipe_ann_lda_smotx, random_grid_lda_quater, n_iter=100, cv=cv, random_state=seed, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 78, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 37.0s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 1.3min\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 2.3min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 2.7min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,...\n", " param_distributions=[{'LDA__n_components': [1, 2, 3],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 78, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_smotx.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 79, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.4096238223269629" ] }, "execution_count": 79, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_smotx.best_score_" ] }, { "cell_type": "code", "execution_count": 80, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'sgd',\n", " 'classifier__hidden_layer_sizes': (10,),\n", " 'classifier__alpha': 0.1,\n", " 'classifier__activation': 'relu',\n", " 'LDA__n_components': 2}" ] }, "execution_count": 80, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_smotx.best_params_" ] }, { "cell_type": "code", "execution_count": 81, "metadata": {}, "outputs": [], "source": [ "param_grid_lda_quater = [{\"LDA__n_components\": [2],\n", " \"classifier__hidden_layer_sizes\": [(10,), (10, 10,), (10, 10, 10,), (10, 10, 10, 10,), (10, 10, 10, 10, 10,)],\n", " \"classifier__activation\": ['relu'],\n", " \"classifier__solver\": ['sgd'],\n", " \"classifier__alpha\": [0.005, 0.01, 0.05, 0.1, 0.5, 1, 5]}]" ] }, { "cell_type": "code", "execution_count": 82, "metadata": {}, "outputs": [], "source": [ "clf_ann_lda_smotx = GridSearchCV(pipe_ann_lda_smotx, param_grid_lda_quater, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 83, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 35 candidates, totalling 175 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 11.0s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 26.9s\n", "[Parallel(n_jobs=-1)]: Done 175 out of 175 | elapsed: 35.0s finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,\n", " tol=0....\n", " iid='deprecated', n_jobs=-1,\n", " param_grid=[{'LDA__n_components': [2],\n", " 'classifier__activation': ['relu'],\n", " 'classifier__alpha': [0.005, 0.01, 0.05, 0.1, 0.5, 1,\n", " 5],\n", " 'classifier__hidden_layer_sizes': [(10,), (10, 10),\n", " (10, 10, 10),\n", " (10, 10, 10, 10),\n", " (10, 10, 10, 10,\n", " 10)],\n", " 'classifier__solver': ['sgd']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 83, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_smotx.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 84, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'LDA__n_components': 2,\n", " 'classifier__activation': 'relu',\n", " 'classifier__alpha': 0.005,\n", " 'classifier__hidden_layer_sizes': (10,),\n", " 'classifier__solver': 'sgd'}" ] }, "execution_count": 84, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_smotx.best_params_" ] }, { "cell_type": "code", "execution_count": 85, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.4096238223269629" ] }, "execution_count": 85, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_smotx.best_score_" ] }, { "cell_type": "code", "execution_count": 86, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4218\n" ] } ], "source": [ "quater_pca_predict = clf_ann_lda_smotx.predict(X_test_quater)\n", "testscore_quater_pca = metrics.f1_score(y_test_quater, quater_pca_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_quater_pca))" ] }, { "cell_type": "code", "execution_count": 87, "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[17 7 0 1]\n", " [ 9 22 3 12]\n", " [ 2 14 3 7]\n", " [ 1 6 3 11]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_quater, quater_pca_predict))" ] }, { "cell_type": "code", "execution_count": 88, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Quaternary_APC_ANN_LDA_smotx.pkl']" ] }, "execution_count": 88, "metadata": {}, "output_type": "execute_result" } ], "source": [ "joblib.dump(clf_ann_lda_smotx, 'Quaternary_APC_ANN_LDA_smotx.pkl')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Quaternary (w/ SMOTE)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "PCA" ] }, { "cell_type": "code", "execution_count": 89, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 32.5s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 1.8min\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 4.8min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 6.5min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True,\n", " iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=Fal...\n", " param_distributions=[{'PCA__n_components': [1, 2, 3],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 89, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 90, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.35272702321954014" ] }, "execution_count": 90, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.best_score_" ] }, { "cell_type": "code", "execution_count": 91, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'adam',\n", " 'classifier__hidden_layer_sizes': (200,),\n", " 'classifier__alpha': 0.001,\n", " 'classifier__activation': 'relu',\n", " 'PCA__n_components': 3}" ] }, "execution_count": 91, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_pca.best_params_" ] }, { "cell_type": "code", "execution_count": 92, "metadata": {}, "outputs": [], "source": [ "param_grid_pca_quater_smot = [{\"PCA__n_components\": [3],\n", " \"classifier__hidden_layer_sizes\": [(200,), (200, 200,), (200, 200, 200,), (200, 200, 200, 200,), (200, 200, 200, 200, 200,)],\n", " \"classifier__activation\": ['relu'],\n", " \"classifier__solver\": ['adam'],\n", " \"classifier__alpha\": [0.00005, 0.0001, 0.0005, 0.001, 0.005, 0.01, 0.05]}]" ] }, { "cell_type": "code", "execution_count": 93, "metadata": {}, "outputs": [], "source": [ "clf_ann_pca_quater_smot = GridSearchCV(pipe_ann_pca, param_grid_pca_quater_smot, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 94, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 35 candidates, totalling 175 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 1.6min\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 8.1min\n", "[Parallel(n_jobs=-1)]: Done 175 out of 175 | elapsed: 9.3min finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('PCA',\n", " PCA(copy=True, iterated_power='auto',\n", " n_components=None,\n", " random_state=None,\n", " svd_solver='auto', tol=0.0,\n", " whiten=False)),\n", " (...\n", " param_grid=[{'PCA__n_components': [3],\n", " 'classifier__activation': ['relu'],\n", " 'classifier__alpha': [5e-05, 0.0001, 0.0005, 0.001,\n", " 0.005, 0.01, 0.05],\n", " 'classifier__hidden_layer_sizes': [(200,), (200, 200),\n", " (200, 200, 200),\n", " (200, 200, 200,\n", " 200),\n", " (200, 200, 200,\n", " 200, 200)],\n", " 'classifier__solver': ['adam']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 94, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_quater_smot.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 95, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'PCA__n_components': 3,\n", " 'classifier__activation': 'relu',\n", " 'classifier__alpha': 0.01,\n", " 'classifier__hidden_layer_sizes': (200, 200),\n", " 'classifier__solver': 'adam'}" ] }, "execution_count": 95, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_quater_smot.best_params_" ] }, { "cell_type": "code", "execution_count": 96, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.37698143449720856" ] }, "execution_count": 96, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_pca_quater_smot.best_score_" ] }, { "cell_type": "code", "execution_count": 97, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4746\n" ] } ], "source": [ "quater_smot_pca_predict = clf_ann_pca_quater_smot.predict(X_test_quater)\n", "testscore_quater_smot_pca = metrics.f1_score(y_test_quater, quater_smot_pca_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_quater_smot_pca))" ] }, { "cell_type": "code", "execution_count": 98, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[15 4 3 3]\n", " [ 2 17 16 11]\n", " [ 2 8 10 6]\n", " [ 0 3 6 12]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_quater, quater_smot_pca_predict))" ] }, { "cell_type": "code", "execution_count": 99, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Quaternary_APC_ANN_PCA_smot.pkl']" ] }, "execution_count": 99, "metadata": {}, "output_type": "execute_result" } ], "source": [ "joblib.dump(clf_ann_pca_quater_smot, 'Quaternary_APC_ANN_PCA_smot.pkl')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "LDA" ] }, { "cell_type": "code", "execution_count": 100, "metadata": {}, "outputs": [], "source": [ "rand_ann_lda_quater = RandomizedSearchCV(pipe_ann_lda, random_grid_lda_quater, n_iter=100, cv=cv, random_state=seed, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 101, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 100 candidates, totalling 500 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 46.7s\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 2.0min\n", "[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 3.3min\n", "[Parallel(n_jobs=-1)]: Done 500 out of 500 | elapsed: 3.9min finished\n" ] }, { "data": { "text/plain": [ "RandomizedSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0,\n", " 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,...\n", " param_distributions=[{'LDA__n_components': [1, 2, 3],\n", " 'classifier__activation': ['relu',\n", " 'logistic',\n", " 'tanh',\n", " 'identity'],\n", " 'classifier__alpha': [1e-05, 0.001,\n", " 0.1, 10],\n", " 'classifier__hidden_layer_sizes': [(10,),\n", " (50,),\n", " (100,),\n", " (200,)],\n", " 'classifier__solver': ['lbfgs', 'adam',\n", " 'sgd']}],\n", " pre_dispatch='2*n_jobs', random_state=28, refit=True,\n", " return_train_score=False, scoring='f1_macro', verbose=2)" ] }, "execution_count": 101, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_quater.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 102, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.407606924055982" ] }, "execution_count": 102, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_quater.best_score_" ] }, { "cell_type": "code", "execution_count": 103, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'classifier__solver': 'adam',\n", " 'classifier__hidden_layer_sizes': (200,),\n", " 'classifier__alpha': 0.001,\n", " 'classifier__activation': 'relu',\n", " 'LDA__n_components': 3}" ] }, "execution_count": 103, "metadata": {}, "output_type": "execute_result" } ], "source": [ "rand_ann_lda_quater.best_params_" ] }, { "cell_type": "code", "execution_count": 104, "metadata": {}, "outputs": [], "source": [ "param_grid_lda_quater_smot = [{\"LDA__n_components\": [3],\n", " \"classifier__hidden_layer_sizes\": [(200,), (200, 200,), (200, 200, 200,), (200, 200, 200, 200,), (200, 200, 200, 200, 200,)],\n", " \"classifier__activation\": ['relu'],\n", " \"classifier__solver\": ['adam'],\n", " \"classifier__alpha\": [0.00005, 0.0001, 0.0005, 0.001, 0.005, 0.01, 0.05]}]" ] }, { "cell_type": "code", "execution_count": 105, "metadata": {}, "outputs": [], "source": [ "clf_ann_lda_quater_smot = GridSearchCV(pipe_ann_lda, param_grid_lda_quater_smot, cv=cv, scoring='f1_macro', verbose=2, n_jobs=-1)" ] }, { "cell_type": "code", "execution_count": 106, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fitting 5 folds for each of 35 candidates, totalling 175 fits\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n", "[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 1.4min\n", "[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 7.7min\n", "[Parallel(n_jobs=-1)]: Done 175 out of 175 | elapsed: 9.0min finished\n" ] }, { "data": { "text/plain": [ "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=28, shuffle=True),\n", " error_score=nan,\n", " estimator=Pipeline(memory=None,\n", " steps=[('std',\n", " MinMaxScaler(copy=True,\n", " feature_range=(0, 1))),\n", " ('thresholder',\n", " VarianceThreshold(threshold=0.01)),\n", " ('LDA',\n", " LinearDiscriminantAnalysis(n_components=None,\n", " priors=None,\n", " shrinkage=None,\n", " solver='svd',\n", " store_covariance=False,\n", " tol=0....\n", " param_grid=[{'LDA__n_components': [3],\n", " 'classifier__activation': ['relu'],\n", " 'classifier__alpha': [5e-05, 0.0001, 0.0005, 0.001,\n", " 0.005, 0.01, 0.05],\n", " 'classifier__hidden_layer_sizes': [(200,), (200, 200),\n", " (200, 200, 200),\n", " (200, 200, 200,\n", " 200),\n", " (200, 200, 200,\n", " 200, 200)],\n", " 'classifier__solver': ['adam']}],\n", " pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n", " scoring='f1_macro', verbose=2)" ] }, "execution_count": 106, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_quater_smot.fit(X_train_quater, y_train_quater.values.ravel())" ] }, { "cell_type": "code", "execution_count": 107, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'LDA__n_components': 3,\n", " 'classifier__activation': 'relu',\n", " 'classifier__alpha': 0.005,\n", " 'classifier__hidden_layer_sizes': (200, 200),\n", " 'classifier__solver': 'adam'}" ] }, "execution_count": 107, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_quater_smot.best_params_" ] }, { "cell_type": "code", "execution_count": 108, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.40966437761403685" ] }, "execution_count": 108, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf_ann_lda_quater_smot.best_score_" ] }, { "cell_type": "code", "execution_count": 109, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "f1 score (test): 0.4152\n" ] } ], "source": [ "quater_smot_lda_predict = clf_ann_lda_quater_smot.predict(X_test_quater)\n", "testscore_quater_smot_lda = metrics.f1_score(y_test_quater, quater_smot_lda_predict, average='macro')\n", "\n", "print(\"f1 score (test): {:.4f}\".format(testscore_quater_smot_lda))" ] }, { "cell_type": "code", "execution_count": 110, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "오차 행렬: \n", " [[16 5 4 0]\n", " [ 9 18 5 14]\n", " [ 1 12 4 9]\n", " [ 0 5 4 12]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "print(\"오차 행렬: \\n\", confusion_matrix(y_test_quater, quater_smot_lda_predict))" ] }, { "cell_type": "code", "execution_count": 111, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Quaternary_APC_ANN_LDA_smot.pkl']" ] }, "execution_count": 111, "metadata": {}, "output_type": "execute_result" } ], "source": [ "joblib.dump(clf_ann_lda_quater_smot, 'Quaternary_APC_ANN_LDA_smot.pkl')" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.4" } }, "nbformat": 4, "nbformat_minor": 2 }