{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "38967358-1993-4d2a-ba1e-bfe4f484cdea", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Period: 2020 - 2024\n", "Rows: 1827\n", "Unique dates: 1827\n", "Date range: 2020-01-01 to 2024-12-31\n", "\\nSaved outputs:\n", " - Figure1_Boxplot_peak_vs_rest_bw.tif\n", " - Figure2_Histograms_raw_log_bw.tif\n", " - Figure3_Season_duration_start_end_bw.tif\n", " - Figure4_Start_DOY_vs_thermal_indices_bw.tif\n", " - Figure5_Spearman_correlation_bw.tif\n", " - Figure6_Scree_plot_bw.tif\n", " - Figure7_PCA_biplot_bw.tif\n", " - Figure8_Varimax_rotation_bw.tif\n", " - NB_full_summary.txt\n", " - NB_season_summary.txt\n", " - Table1_Descriptive_statistics.csv\n", " - Table2_Season_timing.csv\n", " - Table3_Start_DOY_vs_thermal_indices.csv\n", " - Table4_ZINB_results.csv\n", " - Table5_PCA_summary.csv\n", " - Varimax_rotated_loadings.csv\n", " - ZINB_summary.txt\n" ] } ], "source": [ "#!/usr/bin/env python3\n", "# -*- coding: utf-8 -*-\n", "\"\"\"\n", "All-in-one analysis script for:\n", "Ambrosia pollen season dynamics and meteorological drivers in Zagreb (2020–2024)\n", "\n", "Outputs:\n", "- Black-and-white journal-style Figures 1–8\n", "- Table 1: Descriptive statistics\n", "- Table 2: Season timing and totals\n", "- Table 3: Start_DOY vs preseason thermal indices\n", "- Table 4: ZINB coefficients\n", "- Table 5: PCA summary\n", "\n", "Input file expected:\n", " MasterExcel_CLEAN_ENG.csv\n", "\n", "Designed for Springer Aerobiologia (black-and-white friendly output).\n", "\"\"\"\n", "\n", "import os\n", "import warnings\n", "import numpy as np\n", "import pandas as pd\n", "import matplotlib.pyplot as plt\n", "from matplotlib.ticker import MaxNLocator\n", "from sklearn.preprocessing import StandardScaler\n", "from sklearn.decomposition import PCA, FactorAnalysis\n", "import seaborn as sns\n", "import statsmodels.api as sm\n", "from statsmodels.discrete.count_model import NegativeBinomialP, ZeroInflatedNegativeBinomialP\n", "\n", "warnings.filterwarnings(\"ignore\")\n", "\n", "INPUT_FILE = \"MasterExcel_CLEAN_ENG.csv\"\n", "OUTDIR = \"Aerobiologia_bw_outputs\"\n", "os.makedirs(OUTDIR, exist_ok=True)\n", "\n", "plt.rcParams.update({\n", " \"figure.dpi\": 120,\n", " \"savefig.dpi\": 600,\n", " \"font.size\": 10,\n", " \"axes.titlesize\": 11,\n", " \"axes.labelsize\": 10,\n", " \"xtick.labelsize\": 9,\n", " \"ytick.labelsize\": 9,\n", " \"legend.fontsize\": 9,\n", " \"axes.linewidth\": 0.8,\n", " \"lines.linewidth\": 1.2,\n", " \"patch.linewidth\": 0.8,\n", " \"font.family\": \"DejaVu Sans\",\n", "})\n", "\n", "BW = {\n", " \"black\": \"black\",\n", " \"dark\": \"0.25\",\n", " \"mid\": \"0.50\",\n", " \"light\": \"0.75\",\n", " \"verylight\": \"0.90\"\n", "}\n", "\n", "def savefig(name):\n", " plt.tight_layout()\n", " plt.savefig(os.path.join(OUTDIR, name), bbox_inches=\"tight\")\n", " plt.close()\n", "\n", "def exp_or_nan(x):\n", " try:\n", " return np.exp(x)\n", " except Exception:\n", " return np.nan\n", "\n", "def safe_p(p):\n", " if pd.isna(p):\n", " return np.nan\n", " if p < 0.001:\n", " return \"<0.001\"\n", " return round(float(p), 3)\n", "\n", "def varimax(Phi, gamma=1.0, q=100, tol=1e-6):\n", " p, k = Phi.shape\n", " R = np.eye(k)\n", " d = 0\n", " for _ in range(q):\n", " d_old = d\n", " Lambda = np.dot(Phi, R)\n", " u, s, vh = np.linalg.svd(\n", " np.dot(\n", " Phi.T,\n", " np.asarray(Lambda) ** 3 -\n", " (gamma / p) * np.dot(Lambda, np.diag(np.diag(Lambda.T @ Lambda)))\n", " )\n", " )\n", " R = np.dot(u, vh)\n", " d = np.sum(s)\n", " if d_old != 0 and d / d_old < 1 + tol:\n", " break\n", " return np.dot(Phi, R), R\n", "\n", "\n", "# =========================================================\n", "# 1) LOAD DATA\n", "# =========================================================\n", "df = pd.read_csv(INPUT_FILE)\n", "df.columns = df.columns.astype(str).str.strip()\n", "\n", "COLS_EN = {\n", " \"Godina\": \"Year\",\n", " \"Mjesec\": \"Month\",\n", " \"Dan\": \"Day\",\n", " \"Temp\": \"Temperature\",\n", " \"Oborina\": \"Precipitation\",\n", " \"Vlaga\": \"Relative humidity\",\n", " \"Tlak\": \"Pressure\",\n", " \"Brzina_vjetra\": \"Wind speed\",\n", " \"Datum\": \"Date\",\n", " \"Broj zrnaca Breza\": \"Birch pollen\",\n", " \"Broj zrnaca Ambrozija\": \"Ragweed pollen\",\n", " \"Broj zrnaca Trave\": \"Grass pollen\",\n", "}\n", "df = df.rename(columns=COLS_EN)\n", "\n", "required_cols = [\n", " \"Date\", \"Year\", \"Month\", \"Temperature\", \"Precipitation\",\n", " \"Relative humidity\", \"Wind speed\", \"Ragweed pollen\"\n", "]\n", "missing = [c for c in required_cols if c not in df.columns]\n", "if missing:\n", " raise ValueError(f\"Missing required columns: {missing}\")\n", "\n", "df[\"Date\"] = pd.to_datetime(df[\"Date\"], errors=\"coerce\")\n", "if df[\"Date\"].isna().any():\n", " raise ValueError(\"Date parsing produced NA values. Check the Date column.\")\n", "\n", "df = df[(df[\"Year\"] >= 2020) & (df[\"Year\"] <= 2024)].copy()\n", "df = df.sort_values(\"Date\").reset_index(drop=True)\n", "\n", "df[\"Month\"] = pd.to_numeric(df[\"Month\"], errors=\"coerce\")\n", "df.loc[df[\"Month\"].isna(), \"Month\"] = df.loc[df[\"Month\"].isna(), \"Date\"].dt.month\n", "\n", "df[\"log_ragweed\"] = np.log1p(df[\"Ragweed pollen\"])\n", "df[\"Season_group\"] = np.where(df[\"Month\"].isin([8, 9]), \"August–September\", \"Rest of year\")\n", "\n", "print(\"Period:\", int(df[\"Year\"].min()), \"-\", int(df[\"Year\"].max()))\n", "print(\"Rows:\", df.shape[0])\n", "print(\"Unique dates:\", df[\"Date\"].nunique())\n", "print(\"Date range:\", df[\"Date\"].min().date(), \"to\", df[\"Date\"].max().date())\n", "\n", "\n", "# =========================================================\n", "# 2) TABLE 1 — DESCRIPTIVE STATISTICS\n", "# =========================================================\n", "table1 = pd.DataFrame([{\n", " \"N_days\": int(df.shape[0]),\n", " \"N_zero_days\": int((df[\"Ragweed pollen\"] == 0).sum()),\n", " \"Mean\": round(df[\"Ragweed pollen\"].mean(), 2),\n", " \"Median\": round(df[\"Ragweed pollen\"].median(), 2),\n", " \"Std\": round(df[\"Ragweed pollen\"].std(), 2),\n", " \"Min\": round(df[\"Ragweed pollen\"].min(), 2),\n", " \"P25\": round(df[\"Ragweed pollen\"].quantile(0.25), 2),\n", " \"P75\": round(df[\"Ragweed pollen\"].quantile(0.75), 2),\n", " \"P95\": round(df[\"Ragweed pollen\"].quantile(0.95), 2),\n", " \"Max\": round(df[\"Ragweed pollen\"].max(), 2),\n", "}])\n", "table1.to_csv(os.path.join(OUTDIR, \"Table1_Descriptive_statistics.csv\"), index=False)\n", "\n", "\n", "# =========================================================\n", "# 3) FIGURE 1 — BOXPLOT (B/W)\n", "# =========================================================\n", "plt.figure(figsize=(6.4, 5.2))\n", "sns.boxplot(\n", " data=df,\n", " x=\"Season_group\",\n", " y=\"Ragweed pollen\",\n", " color=\"white\",\n", " width=0.5,\n", " showcaps=True,\n", " boxprops=dict(edgecolor=\"black\", facecolor=\"white\"),\n", " whiskerprops=dict(color=\"black\"),\n", " capprops=dict(color=\"black\"),\n", " medianprops=dict(color=\"black\", linewidth=1.2),\n", " flierprops=dict(marker='o', markerfacecolor='white', markeredgecolor='black',\n", " markersize=3, linestyle='none'),\n", ")\n", "plt.yscale(\"log\")\n", "plt.xlabel(\"\")\n", "plt.ylabel(\"Ragweed pollen (grains m$^{-3}$, log scale)\")\n", "plt.title(\"Peak flowering period vs. rest of the year\")\n", "savefig(\"Figure1_Boxplot_peak_vs_rest_bw.tif\")\n", "\n", "\n", "# =========================================================\n", "# 4) FIGURE 2 — HISTOGRAMS (B/W)\n", "# =========================================================\n", "fig, axes = plt.subplots(1, 2, figsize=(10, 4.3))\n", "\n", "axes[0].hist(df[\"Ragweed pollen\"].dropna(), bins=40, color=\"white\", edgecolor=\"black\")\n", "axes[0].set_title(\"(a) Daily values\")\n", "axes[0].set_xlabel(\"Ragweed pollen (grains m$^{-3}$)\")\n", "axes[0].set_ylabel(\"Frequency\")\n", "\n", "axes[1].hist(df[\"log_ragweed\"].dropna(), bins=40, color=BW[\"light\"], edgecolor=\"black\")\n", "axes[1].set_title(\"(b) log(1 + x)\")\n", "axes[1].set_xlabel(\"log(Ragweed pollen + 1)\")\n", "axes[1].set_ylabel(\"Frequency\")\n", "\n", "fig.suptitle(\"Histograms of ragweed pollen concentrations (2020–2024)\", y=1.02)\n", "plt.tight_layout()\n", "plt.savefig(os.path.join(OUTDIR, \"Figure2_Histograms_raw_log_bw.tif\"), bbox_inches=\"tight\", dpi=600)\n", "plt.close()\n", "\n", "\n", "# =========================================================\n", "# 5) SEASON METRICS — TABLES 2 AND 3\n", "# =========================================================\n", "season_rows = []\n", "for year, g in df.groupby(\"Year\"):\n", " g = g.sort_values(\"Date\").copy()\n", " total = g[\"Ragweed pollen\"].sum()\n", " if total <= 0:\n", " continue\n", "\n", " g[\"cum_sum\"] = g[\"Ragweed pollen\"].cumsum()\n", " g[\"cum_pct\"] = g[\"cum_sum\"] / total * 100\n", "\n", " start_row = g[g[\"cum_pct\"] >= 5].iloc[0]\n", " end_row = g[g[\"cum_pct\"] >= 95].iloc[0]\n", "\n", " start_date = start_row[\"Date\"]\n", " end_date = end_row[\"Date\"]\n", "\n", " g_prev = g[g[\"Date\"] < start_date].copy()\n", " if not g_prev.empty:\n", " g_prev[\"gdd10\"] = np.maximum(g_prev[\"Temperature\"] - 10, 0)\n", " gdd10_prev = g_prev[\"gdd10\"].sum()\n", " else:\n", " gdd10_prev = np.nan\n", "\n", " t30_window = df[(df[\"Date\"] >= start_date - pd.Timedelta(days=30)) & (df[\"Date\"] < start_date)]\n", " t30_prev = t30_window[\"Temperature\"].mean()\n", "\n", " season_rows.append({\n", " \"Year\": int(year),\n", " \"Start\": start_date.date().isoformat(),\n", " \"End\": end_date.date().isoformat(),\n", " \"Duration\": int((end_date - start_date).days),\n", " \"Ambrosia pollen count/m3\": int(total),\n", " \"Start_DOY\": int(start_date.dayofyear),\n", " \"End_DOY\": int(end_date.dayofyear),\n", " \"GDD10_prev\": round(gdd10_prev, 1),\n", " \"MeanT30_prev\": round(t30_prev, 2)\n", " })\n", "\n", "season_df = pd.DataFrame(season_rows).sort_values(\"Year\")\n", "\n", "table2 = season_df[[\"Year\", \"Start\", \"End\", \"Duration\", \"Ambrosia pollen count/m3\"]].copy()\n", "table2.to_csv(os.path.join(OUTDIR, \"Table2_Season_timing.csv\"), index=False)\n", "\n", "table3 = season_df[[\"Year\", \"Start_DOY\", \"GDD10_prev\", \"MeanT30_prev\"]].copy()\n", "table3.to_csv(os.path.join(OUTDIR, \"Table3_Start_DOY_vs_thermal_indices.csv\"), index=False)\n", "\n", "\n", "# =========================================================\n", "# 6) FIGURE 3 — SEASON CHARACTERISTICS (B/W)\n", "# =========================================================\n", "fig, axes = plt.subplots(1, 2, figsize=(10.2, 4.4))\n", "\n", "axes[0].bar(season_df[\"Year\"].astype(str), season_df[\"Duration\"],\n", " color=\"white\", edgecolor=\"black\", hatch=\"///\")\n", "axes[0].set_title(\"(a) Season duration\")\n", "axes[0].set_xlabel(\"Year\")\n", "axes[0].set_ylabel(\"Duration (days)\")\n", "\n", "axes[1].plot(season_df[\"Year\"], season_df[\"Start_DOY\"],\n", " marker=\"o\", linestyle=\"-\", color=\"black\", label=\"Start\")\n", "axes[1].plot(season_df[\"Year\"], season_df[\"End_DOY\"],\n", " marker=\"s\", linestyle=\"--\", color=BW[\"dark\"], label=\"End\")\n", "axes[1].set_title(\"(b) Start and end of season\")\n", "axes[1].set_xlabel(\"Year\")\n", "axes[1].set_ylabel(\"Day of year (DOY)\")\n", "axes[1].legend(frameon=False)\n", "\n", "fig.suptitle(\"Season characteristics of ragweed pollen (2020–2024)\", y=1.02)\n", "plt.tight_layout()\n", "plt.savefig(os.path.join(OUTDIR, \"Figure3_Season_duration_start_end_bw.tif\"), bbox_inches=\"tight\", dpi=600)\n", "plt.close()\n", "\n", "\n", "# =========================================================\n", "# 7) FIGURE 4 — START_DOY VS THERMAL INDICES (B/W)\n", "# =========================================================\n", "fig, axes = plt.subplots(1, 2, figsize=(10.2, 4.4))\n", "\n", "axes[0].scatter(season_df[\"GDD10_prev\"], season_df[\"Start_DOY\"],\n", " s=40, facecolors=\"white\", edgecolors=\"black\")\n", "for _, row in season_df.iterrows():\n", " axes[0].text(row[\"GDD10_prev\"], row[\"Start_DOY\"], str(int(row[\"Year\"])),\n", " fontsize=8, ha=\"left\", va=\"bottom\")\n", "axes[0].set_xlabel(\"GDD10 before season onset\")\n", "axes[0].set_ylabel(\"Start of season (DOY)\")\n", "axes[0].set_title(\"(a) GDD10\")\n", "\n", "axes[1].scatter(season_df[\"MeanT30_prev\"], season_df[\"Start_DOY\"],\n", " s=40, facecolors=\"white\", edgecolors=\"black\")\n", "for _, row in season_df.iterrows():\n", " axes[1].text(row[\"MeanT30_prev\"], row[\"Start_DOY\"], str(int(row[\"Year\"])),\n", " fontsize=8, ha=\"left\", va=\"bottom\")\n", "axes[1].set_xlabel(\"Mean temperature in previous 30 days (°C)\")\n", "axes[1].set_ylabel(\"Start of season (DOY)\")\n", "axes[1].set_title(\"(b) T-30\")\n", "\n", "fig.suptitle(\"Start-of-season DOY versus preseason thermal indices\", y=1.02)\n", "plt.tight_layout()\n", "plt.savefig(os.path.join(OUTDIR, \"Figure4_Start_DOY_vs_thermal_indices_bw.tif\"), bbox_inches=\"tight\", dpi=600)\n", "plt.close()\n", "\n", "\n", "# =========================================================\n", "# 8) FIGURE 5 — SPEARMAN CORRELATION MATRIX (GRAYSCALE)\n", "# =========================================================\n", "corr_vars = [\"Ragweed pollen\", \"Temperature\", \"Relative humidity\", \"Precipitation\", \"Wind speed\"]\n", "corr = df[corr_vars].corr(method=\"spearman\")\n", "\n", "labels = {\n", " \"Ragweed pollen\": \"Ragweed pollen\",\n", " \"Temperature\": \"Temperature\",\n", " \"Relative humidity\": \"Humidity\",\n", " \"Precipitation\": \"Precipitation\",\n", " \"Wind speed\": \"Wind speed\",\n", "}\n", "corr_plot = corr.rename(index=labels, columns=labels)\n", "\n", "plt.figure(figsize=(6.8, 5.8))\n", "sns.heatmap(\n", " corr_plot,\n", " annot=True,\n", " fmt=\".2f\",\n", " cmap=\"Greys\",\n", " vmin=-1,\n", " vmax=1,\n", " square=True,\n", " linewidths=0.6,\n", " linecolor=\"white\",\n", " cbar_kws={\"label\": \"Spearman correlation\", \"shrink\": 0.9},\n", " annot_kws={\"size\": 9, \"color\": \"black\"}\n", ")\n", "plt.title(\"Spearman correlation matrix\", pad=10)\n", "plt.xticks(rotation=30, ha=\"right\")\n", "plt.yticks(rotation=0)\n", "savefig(\"Figure5_Spearman_correlation_bw.tif\")\n", "\n", "\n", "# =========================================================\n", "# 9) NB + ZINB MODELS\n", "# =========================================================\n", "df_model = df[[\n", " \"Date\", \"Month\", \"Ragweed pollen\", \"Temperature\",\n", " \"Relative humidity\", \"Precipitation\", \"Wind speed\"\n", "]].copy()\n", "\n", "doy = df_model[\"Date\"].dt.dayofyear\n", "df_model[\"sin_doy\"] = np.sin(2 * np.pi * doy / 365)\n", "df_model[\"cos_doy\"] = np.cos(2 * np.pi * doy / 365)\n", "df_model[\"Season\"] = df_model[\"Date\"].dt.month.isin([8, 9]).astype(int)\n", "df_model[\"pollen_lag1\"] = df_model[\"Ragweed pollen\"].shift(1)\n", "df_model = df_model.dropna().copy()\n", "\n", "for src, dst in [\n", " (\"Temperature\", \"Temperature_z\"),\n", " (\"Relative humidity\", \"Humidity_z\"),\n", " (\"Precipitation\", \"Precipitation_z\"),\n", " (\"pollen_lag1\", \"Lag1_z\"),\n", "]:\n", " df_model[dst] = (df_model[src] - df_model[src].mean()) / df_model[src].std()\n", "\n", "# Full-year NB\n", "vars_nb_full = [\"Temperature_z\", \"Humidity_z\", \"Precipitation_z\", \"Wind speed\",\n", " \"sin_doy\", \"cos_doy\", \"Season\"]\n", "X_nb_full = sm.add_constant(df_model[vars_nb_full])\n", "y_nb_full = df_model[\"Ragweed pollen\"].astype(int)\n", "nb_full = NegativeBinomialP(endog=y_nb_full, exog=X_nb_full).fit(disp=False, maxiter=200)\n", "\n", "# Season-restricted NB\n", "df_season_nb = df_model[df_model[\"Season\"] == 1].copy()\n", "vars_nb_season = [\"Temperature_z\", \"Humidity_z\", \"Precipitation_z\", \"Wind speed\"]\n", "X_nb_season = sm.add_constant(df_season_nb[vars_nb_season])\n", "y_nb_season = df_season_nb[\"Ragweed pollen\"].astype(int)\n", "nb_season = NegativeBinomialP(endog=y_nb_season, exog=X_nb_season).fit(disp=False, maxiter=200)\n", "\n", "# Final ZINB\n", "vars_count = [\"Temperature_z\", \"Lag1_z\", \"sin_doy\", \"cos_doy\", \"Season\"]\n", "vars_infl = [\"Temperature_z\", \"Lag1_z\"]\n", "\n", "X = sm.add_constant(df_model[vars_count])\n", "Z = sm.add_constant(df_model[vars_infl])\n", "y = df_model[\"Ragweed pollen\"].astype(int)\n", "\n", "zinb = ZeroInflatedNegativeBinomialP(\n", " endog=y,\n", " exog=X,\n", " exog_infl=Z,\n", " inflation=\"logit\"\n", ").fit(maxiter=200, method=\"lbfgs\", disp=False)\n", "\n", "\n", "# =========================================================\n", "# 10) TABLE 4 — ZINB COEFFICIENTS\n", "# =========================================================\n", "params = zinb.params\n", "bse = zinb.bse\n", "pvals = zinb.pvalues\n", "conf = zinb.conf_int()\n", "conf.columns = [\"CI_low_coef\", \"CI_high_coef\"]\n", "\n", "rows = []\n", "for term in params.index:\n", " coef = params[term]\n", " se = bse[term]\n", " p = pvals[term]\n", " low = conf.loc[term, \"CI_low_coef\"]\n", " high = conf.loc[term, \"CI_high_coef\"]\n", "\n", " if term == \"alpha\":\n", " rows.append({\n", " \"Predictor\": \"α (dispersion)\",\n", " \"Model component\": \"Overdispersion parameter\",\n", " \"Coefficient\": round(coef, 3),\n", " \"SE\": round(se, 3),\n", " \"IRR/OR\": \"–\",\n", " \"95% CI\": f\"{low:.3f} – {high:.3f}\",\n", " \"p-value\": safe_p(p)\n", " })\n", " continue\n", "\n", " if term.startswith(\"inflate_\"):\n", " component = \"Zero inflation (logit)\"\n", " predictor = term.replace(\"inflate_\", \"\")\n", " predictor = {\n", " \"const\": \"Intercept\",\n", " \"Temperature_z\": \"Temperature (z)\",\n", " \"Lag1_z\": \"Lagged pollen (t−1)\",\n", " }.get(predictor, predictor)\n", "\n", " ratio = exp_or_nan(coef)\n", " low_r = exp_or_nan(low)\n", " high_r = exp_or_nan(high)\n", " ci_txt = f\"{low_r:.3f} – {high_r:.3f}\"\n", "\n", " else:\n", " component = \"Count model (NB)\"\n", " predictor = {\n", " \"const\": \"Intercept\",\n", " \"Temperature_z\": \"Temperature (z)\",\n", " \"Lag1_z\": \"Lagged pollen (t−1)\",\n", " \"sin_doy\": \"sin(DOY)\",\n", " \"cos_doy\": \"cos(DOY)\",\n", " \"Season\": \"Season (Aug–Sep)\"\n", " }.get(term, term)\n", "\n", " if predictor in [\"Intercept\", \"sin(DOY)\"]:\n", " ratio = np.nan\n", " ci_txt = f\"{low:.3f} – {high:.3f}\"\n", " else:\n", " ratio = exp_or_nan(coef)\n", " low_r = exp_or_nan(low)\n", " high_r = exp_or_nan(high)\n", " ci_txt = f\"{low_r:.3f} – {high_r:.3f}\"\n", "\n", " ratio_txt = \"–\" if pd.isna(ratio) else f\"{ratio:.3f}\"\n", " rows.append({\n", " \"Predictor\": predictor,\n", " \"Model component\": component,\n", " \"Coefficient\": round(coef, 3),\n", " \"SE\": round(se, 3),\n", " \"IRR/OR\": ratio_txt,\n", " \"95% CI\": ci_txt,\n", " \"p-value\": safe_p(p)\n", " })\n", "\n", "table4 = pd.DataFrame(rows)\n", "table4.to_csv(os.path.join(OUTDIR, \"Table4_ZINB_results.csv\"), index=False)\n", "\n", "\n", "# =========================================================\n", "# 11) AUG–SEP PCA / EFA\n", "# =========================================================\n", "df_mult = df[df[\"Month\"].isin([8, 9])].copy()\n", "vars_mult = [\"Temperature\", \"Relative humidity\", \"Precipitation\", \"Wind speed\", \"log_ragweed\"]\n", "X_mult = df_mult[vars_mult].dropna().copy()\n", "X_scaled = StandardScaler().fit_transform(X_mult)\n", "\n", "plot_names = {\n", " \"Temperature\": \"Temperature\",\n", " \"Relative humidity\": \"Humidity\",\n", " \"Precipitation\": \"Precipitation\",\n", " \"Wind speed\": \"Wind speed\",\n", " \"log_ragweed\": \"log(Ragweed pollen + 1)\"\n", "}\n", "\n", "# Table 5\n", "pca_full = PCA()\n", "pca_full.fit(X_scaled)\n", "eig = pca_full.explained_variance_\n", "var_pct = pca_full.explained_variance_ratio_ * 100\n", "cum_pct = np.cumsum(var_pct)\n", "\n", "table5 = pd.DataFrame({\n", " \"PC\": [f\"PC{i}\" for i in range(1, len(eig)+1)],\n", " \"Eigenvalue\": np.round(eig, 2),\n", " \"%Var\": np.round(var_pct, 1),\n", " \"Cum%Var\": np.round(cum_pct, 1)\n", "})\n", "table5.to_csv(os.path.join(OUTDIR, \"Table5_PCA_summary.csv\"), index=False)\n", "\n", "\n", "# =========================================================\n", "# 12) FIGURE 6 — SCREE PLOT\n", "# =========================================================\n", "components = np.arange(1, len(eig) + 1)\n", "\n", "plt.figure(figsize=(6.4, 5.0))\n", "plt.plot(components, eig, marker=\"o\", color=\"black\", linestyle=\"-\")\n", "plt.axhline(1, linestyle=\"--\", color=BW[\"dark\"], linewidth=1.0)\n", "plt.xlabel(\"Principal component\")\n", "plt.ylabel(\"Eigenvalue\")\n", "plt.title(\"Scree plot with Kaiser criterion\")\n", "plt.xticks(components)\n", "plt.gca().xaxis.set_major_locator(MaxNLocator(integer=True))\n", "savefig(\"Figure6_Scree_plot_bw.tif\")\n", "\n", "\n", "# =========================================================\n", "# 13) FIGURE 7 — PCA BIPLOT\n", "# =========================================================\n", "pca = PCA(n_components=2)\n", "scores = pca.fit_transform(X_scaled)\n", "raw_loadings = pca.components_.T\n", "\n", "pc1_var = pca.explained_variance_ratio_[0] * 100\n", "pc2_var = pca.explained_variance_ratio_[1] * 100\n", "\n", "temp_idx = vars_mult.index(\"Temperature\")\n", "if raw_loadings[temp_idx, 0] < 0:\n", " scores[:, 0] *= -1\n", " raw_loadings[:, 0] *= -1\n", "\n", "plt.figure(figsize=(7.0, 5.6))\n", "plt.scatter(scores[:, 0], scores[:, 1], s=20, alpha=0.35,\n", " facecolors=\"white\", edgecolors=\"black\")\n", "plt.axhline(0, linewidth=0.8, color=BW[\"dark\"])\n", "plt.axvline(0, linewidth=0.8, color=BW[\"dark\"])\n", "\n", "arrow_scale = 2.8\n", "for i, var in enumerate(vars_mult):\n", " x_arrow = raw_loadings[i, 0] * arrow_scale\n", " y_arrow = raw_loadings[i, 1] * arrow_scale\n", " plt.arrow(\n", " 0, 0, x_arrow, y_arrow,\n", " color=\"black\", linewidth=1.0,\n", " head_width=0.06, head_length=0.09,\n", " length_includes_head=True\n", " )\n", " x_text = x_arrow * 1.10\n", " y_text = y_arrow * 1.10\n", " ha = \"left\" if x_text >= 0 else \"right\"\n", " va = \"bottom\" if y_text >= 0 else \"top\"\n", " plt.text(x_text, y_text, plot_names[var], ha=ha, va=va, fontsize=9)\n", "\n", "plt.xlabel(f\"PC1 ({pc1_var:.1f}% variance)\")\n", "plt.ylabel(f\"PC2 ({pc2_var:.1f}% variance)\")\n", "plt.title(\"PCA biplot (PC1 vs PC2) – August–September\")\n", "savefig(\"Figure7_PCA_biplot_bw.tif\")\n", "\n", "\n", "# =========================================================\n", "# 14) FIGURE 8 — VARIMAX\n", "# =========================================================\n", "fa = FactorAnalysis(n_components=2, random_state=42)\n", "fa.fit(X_scaled)\n", "\n", "raw_loadings_fa = fa.components_.T\n", "rotated_loadings, _ = varimax(raw_loadings_fa)\n", "\n", "rot_df = pd.DataFrame(\n", " rotated_loadings,\n", " index=[plot_names[v] for v in vars_mult],\n", " columns=[\"Factor 1\", \"Factor 2\"]\n", ")\n", "rot_df.to_csv(os.path.join(OUTDIR, \"Varimax_rotated_loadings.csv\"), index=True)\n", "\n", "plt.figure(figsize=(6.4, 5.6))\n", "plt.scatter(rot_df[\"Factor 1\"], rot_df[\"Factor 2\"], s=30,\n", " facecolors=\"white\", edgecolors=\"black\")\n", "plt.axhline(0, linewidth=0.8, color=BW[\"dark\"])\n", "plt.axvline(0, linewidth=0.8, color=BW[\"dark\"])\n", "\n", "for label, row in rot_df.iterrows():\n", " x, y = row[\"Factor 1\"], row[\"Factor 2\"]\n", " ha = \"left\" if x >= 0 else \"right\"\n", " va = \"bottom\" if y >= 0 else \"top\"\n", " xt, yt = x * 1.06, y * 1.06\n", " if label == \"Wind speed\":\n", " yt -= 0.04\n", " plt.text(xt, yt, label, ha=ha, va=va, fontsize=9)\n", "\n", "plt.xlabel(\"Factor 1\")\n", "plt.ylabel(\"Factor 2\")\n", "plt.title(\"Varimax rotation (two-factor solution)\")\n", "savefig(\"Figure8_Varimax_rotation_bw.tif\")\n", "\n", "\n", "# =========================================================\n", "# 15) SAVE MODEL SUMMARIES\n", "# =========================================================\n", "with open(os.path.join(OUTDIR, \"NB_full_summary.txt\"), \"w\", encoding=\"utf-8\") as f:\n", " f.write(nb_full.summary().as_text())\n", "\n", "with open(os.path.join(OUTDIR, \"NB_season_summary.txt\"), \"w\", encoding=\"utf-8\") as f:\n", " f.write(nb_season.summary().as_text())\n", "\n", "with open(os.path.join(OUTDIR, \"ZINB_summary.txt\"), \"w\", encoding=\"utf-8\") as f:\n", " f.write(zinb.summary().as_text())\n", "\n", "print(\"\\\\nSaved outputs:\")\n", "for fn in sorted(os.listdir(OUTDIR)):\n", " print(\" -\", fn)" ] }, { "cell_type": "code", "execution_count": null, "id": "37daa5b3-3d3b-43fa-ac21-3d3964315ec7", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "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": "3.11.14" } }, "nbformat": 4, "nbformat_minor": 5 }