Logistic_Regression/NominalLogistic_Template.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Last Modified on Sep 20, 2025
@purpose:  This Code produces a stratified random sample of a large
           dataset with a nominal target and a nominal predictor
@notes: Uses sci-kit learn train_test_split with stratify option
@data:  CellphoneActivity.csv with n=165,633 cases and k=19 columns
        Saves CellphoneActivity_StratifiedRS.csv with n=33127 (20% of full data)
@email:  ejones@tamu.edu
"""
# ANSI color codes - use high intensity bright red for maximum visibility on black background
RED   = "\033[38;5;197m"  # Bright 256-color red
GOLD  = "\033[38;5;185m"
TEAL  = "\033[38;5;50m"
GREEN = "\033[38;5;82m"
RESET = "\033[0m"

from AdvancedAnalytics.ReplaceImputeEncode import DT, ReplaceImputeEncode
from sklearn.model_selection import train_test_split
import warnings
import pandas as pd

warnings.filterwarnings("ignore", 
                        category=UserWarning)
data_map = {
    'activity':[DT.Nominal, ('sitting', 'sittingdown',
                            'standing', 'standingup', 'walking')],
    'user':    [DT.Nominal,('debora', 'jose_carlos', 'katia', 'wallace')],
    'gender':  [DT.Binary, ('Man', 'Woman')],
    'age':     [DT.Interval,(20, 80)],
    'height':  [DT.Interval,(1.5, 1.75)],
    'weight':  [DT.Interval,(50, 85)],
    'BMI':     [DT.Interval,(20, 30)],
    'x1':      [DT.Interval,(-750, +750)],
    'y1':      [DT.Interval,(-750, +750)],
    'z1':      [DT.Interval,(-750, +750)],
    'x2':      [DT.Interval,(-750, +750)],
    'y2':      [DT.Interval,(-750, +750)],
    'z2':      [DT.Interval,(-750, +750)],
    'x3':      [DT.Interval,(-750, +750)],
    'y3':      [DT.Interval,(-750, +750)],
    'z3':      [DT.Interval,(-750, +750)],
    'x4':      [DT.Interval,(-750, +750)],
    'y4':      [DT.Interval,(-750, +750)],
    'z4':      [DT.Interval,(-750, +750)]
}
target    = "activity"  # A Nominal target
check_var = ["weight", "x1", "y1", "z1"]  # Interval Check Variables
size      = 0.2        # Sample Size
file_in   = "../data/CellphoneActivity.csv"
file_out  = "../data/CellphoneActivity_StratifiedRS.csv" 
#--------- Create Random Stratified Sample -----------------------------------
categorical_list = []
for key in data_map:
    if data_map[key][0] == DT.Nominal or data_map[key][0] == DT.Binary:
        categorical_list.append(key)
l = len(file_in)
print("")
print(f"{RED}--> READING ORIGINAL DATASET:        {file_in} {RESET}")
df  = pd.read_csv(file_in, index_col=None)

Xt, Xv = train_test_split(df, train_size = size, stratify=df[categorical_list], 
                          random_state=12345)
print(f"{RED}--> SAVING RANDOM STRATIFIED SAMPLE: {file_out} {RESET}\n")
Xt.to_csv(file_out)

df2 = pd.read_csv(file_out)
target2 = df2[target].value_counts()
target1 = df[target].value_counts()
print(f"{TEAL}"+39*"="+f"{RESET}")
print(f"{GREEN}"+2*"*", "COMPARING ORIGINAL TO SAMPLE DATA", 
      2*"*"+f"{RESET}")
print(f"{TEAL}"+39*"="+f"{RESET}\n")

print(f"{TEAL}"+36*"="+f"{RESET}")
print(f"{GREEN}SAMPLE SIZES{RESET}")
print(f"{GREEN}TARGET            ORI    OUT   RATIO{RESET}")
print(f"{TEAL}"+36*"-"+f"{RESET}")
for target_val in target1.index:
    ori    = target1[target_val]
    out    = target2[target_val]
    ratio  = ori/out
    print(f"{target_val:15s}{ori: 7.0f}{out: 7.0f}{ratio: 7.1f}")

for var in check_var:
    print("")
    check_ori = df.groupby (target)[var].mean()
    check_out = df2.groupby(target)[var].mean()
    print(f"{TEAL}"+36*"="+f"{RESET}")
    print(f"{GREEN}CHECK AVERAGES FOR {var.upper():5s}{RESET}")
    print(f"{GREEN}TARGET            ORI    OUT   RATIO{RESET}")
    print(f"{TEAL}"+36*"-"+f"{RESET}")
    for target_val in check_ori.index:
        ori    = check_ori[target_val]
        out    = check_out[target_val]
        ratio  = ori/out
        print(f"{target_val:15s}{ori: 7.2f}{out: 7.2f}{ratio: 7.2f}")