feat: implement logistic regression pipeline with automated JASP-formatted report generation and data inspection utilities

2026-07-09 11:29:27 +07:00
parent 1720207f82
commit d32476ef94
8 changed files with 236 additions and 25 deletions
@@ -0,0 +1,102 @@
+import sys
+sys.stdout.reconfigure(encoding='utf-8')
+import pandas as pd
+import numpy as np
+import openpyxl
+from openpyxl.utils.dataframe import dataframe_to_rows
+from openpyxl.styles import Font, Alignment
+import os
+import shutil
+
+base_dir = r'c:\Users\NASPC\Documents\Du án tại SG tháng 8'
+orig_file = os.path.join(base_dir, 'KẾT QUẢ PHÂN TÍCH.xlsx')
+final_file = os.path.join(base_dir, 'Project_Code_and_Results', 'KẾT_QUẢ_PHÂN_TÍCH_MASTER.xlsx')
+
+# Copy original to new file
+shutil.copy(orig_file, final_file)
+
+# Load workbook
+wb = openpyxl.load_workbook(final_file)
+
+# Helper to process CSV into JASP layout dataframe
+def get_jasp_df(filepath, model_name="M₁"):
+    df = pd.read_csv(filepath, index_col=0)
+    se_col = 'S.E.' if 'S.E.' in df.columns else ('std err' if 'std err' in df.columns else 'SE')
+    
+    estimate = df['Beta (B)']
+    se = df[se_col]
+    odds_ratio = df['Odds Ratio EXP(B)']
+    p_val = df['P-value']
+    
+    z_stat = estimate / se
+    wald = z_stat ** 2
+    lower = np.exp(estimate - 1.96 * se)
+    upper = np.exp(estimate + 1.96 * se)
+    
+    def map_name(name):
+        if name == 'const': return '(Intercept)'
+        if '_' in name:
+            parts = name.split('_')
+            return f"{parts[0]} ({parts[1]})"
+        return name
+    
+    new_index = [map_name(str(i)) for i in df.index]
+    
+    out_df = pd.DataFrame({
+        'Model': [model_name] + [np.nan] * (len(df) - 1),
+        '': new_index,
+        'Estimate': estimate.values,
+        'Standard Error': se.values,
+        'Odds Ratio': odds_ratio.values,
+        'z': z_stat.values,
+        'Wald Statistic': wald.values,
+        'df': [1] * len(df),
+        'p': p_val.values,
+        'Lower bound': lower.values,
+        'Upper bound': upper.values
+    })
+    
+    # Format
+    out_df['p'] = out_df['p'].apply(lambda x: '< .00001' if pd.notnull(x) and x < 0.00001 else (round(x, 5) if pd.notnull(x) else x))
+    out_df = out_df.round(5)
+    return out_df
+
+# Replace a sheet's content
+def replace_sheet(sheet_name, csv_path):
+    idx = wb.sheetnames.index(sheet_name)
+    del wb[sheet_name]
+    ws = wb.create_sheet(sheet_name, idx)
+            
+    # Write Title
+    ws.cell(row=1, column=1, value="Logistic Regression (Updated with Firth/Optimized)").font = Font(bold=True)
+    ws.cell(row=3, column=1, value="Coefficients").font = Font(bold=True)
+    
+    # Headers
+    df_jasp = get_jasp_df(csv_path)
+    headers = list(df_jasp.columns)
+    
+    for c_idx, col_name in enumerate(headers, 1):
+        cell = ws.cell(row=4, column=c_idx, value=col_name)
+        cell.font = Font(bold=True)
+        cell.alignment = Alignment(horizontal='center')
+        
+    ws.cell(row=3, column=10, value="95% Confidence interval").font = Font(bold=True)
+    ws.cell(row=3, column=10).alignment = Alignment(horizontal='center')
+    
+    # Write Data
+    for r_idx, row in enumerate(dataframe_to_rows(df_jasp, index=False, header=False), 5):
+        for c_idx, value in enumerate(row, 1):
+            # Check for NaN float
+            if isinstance(value, float) and np.isnan(value):
+                ws.cell(row=r_idx, column=c_idx, value="")
+            else:
+                ws.cell(row=r_idx, column=c_idx, value=value)
+
+path_don = os.path.join(base_dir, 'Project_Code_and_Results', '4_Logistic_Regression', 'Logistic_Results_Donation_Firth.csv')
+path_dec = os.path.join(base_dir, 'Project_Code_and_Results', '4_Logistic_Regression', 'Logistic_Results_Decision_Final.csv')
+
+replace_sheet('Donation', path_don)
+replace_sheet('Decision', path_dec)
+
+wb.save(final_file)
+print("Tạo thành công MASTER file!")