FEAT: Adding the basic structure of the genetic_algorithm.py

Author: BrenoFariasdaSilva

genetic_algorithm.py

@@ -0,0 +1,128 @@
+"""
+================================================================================
+Genetic Algorithm Feature Selection & Analysis
+================================================================================
+Author      : Breno Farias da Silva
+Created     : 2025-10-07
+Description :
+   This script runs a DEAP-based Genetic Algorithm (GA) to perform feature
+   selection for classification problems. It provides an end-to-end pipeline:
+   dataset loading and cleaning, scaling, GA setup and execution, candidate
+   evaluation (with a Random Forest base estimator), and post-hoc analysis
+   (RFE ranking correlation, CSV summaries and boxplot visualizations).
+
+   Key features include:
+      - DEAP-based GA for binary-mask feature selection
+      - Fitness evaluation using a RandomForest and returning multi-metrics
+        (accuracy, precision, recall, F1, FPR, FNR, elapsed time)
+      - Population sweep support (run GA over a range of population sizes)
+      - Integration with previously-computed RFE rankings for cross-checking
+      - Exports: best-subset text file, CSV summaries and per-feature boxplots
+      - Progress bars via tqdm and safe filename handling for outputs
+      - Cross-platform completion notification (optional sound)
+
+Usage:
+   1. Configure the dataset:
+      - Edit the `csv_file` variable in the `main()` function to point to
+         the CSV dataset you want to analyze (the script assumes the last
+         column is the target and numeric features are used).
+   2. Optionally tune GA parameters in `main()` or call sites:
+      - n_generations, min_pop, max_pop, population_size, train_test_ratio
+   3. Run the pipeline via the project's Makefile:
+      $ make main
+      (Makefile is expected to setup env / deps and execute this script.)
+   NOTE:
+      - If you prefer not to use the Makefile, you can run the module/script
+        directly from Python in your dev environment, but the recommended
+        workflow for the project is `make main`.
+
+Outputs:
+   - Feature_Analysis/Genetic_Algorithm_results.txt  (best subset + RFE cross-info)
+   - Feature_Analysis/<dataset>_feature_summary.csv  (mean/std per class for selected features)
+   - Feature_Analysis/<dataset>-<feature>.png         (boxplots for top features)
+   - Console summary of best subsets per population size (when sweeping)
+
+TODOs:
+   - Add CLI argument parsing (argparse) to avoid editing `main()` for different runs.
+   - Add cross-validation or nested CV to make fitness evaluation more robust.
+   - Support multi-objective optimization (e.g., F1 vs. model training time).
+   - Parallelize individual evaluations (joblib / dask) to speed up GA fitness calls.
+   - Save and version best individuals (pickle/JSON) and GA run metadata.
+   - Implement reproducible seeding across DEAP, numpy, random and sklearn.
+   - Add automatic handling of categorical features and missing-value imputation.
+   - Add early stopping and convergence checks to the GA loop.
+   - Produce a machine-readable summary (JSON) of final metrics and selected features.
+   - Add unit tests for core functions (fitness evaluation, GA setup, I/O).
+
+Dependencies:
+   - Python >= 3.9
+   - pandas, numpy, scikit-learn, deap, tqdm, matplotlib, seaborn, colorama
+
+Assumptions & Notes:
+   - Dataset format: CSV, last column = target. Only numeric features are used.
+   - RFE results (if present) are read from `Feature_Analysis/RFE_results_RandomForestClassifier.txt`.
+   - Sound notification is skipped on Windows by default.
+   - The script uses RandomForestClassifier as the default evaluator; change as needed.
+   - Inspect output directories (`Feature_Analysis/`) after runs for artifacts.
+"""
+
+import atexit # For playing a sound when the program finishes
+import matplotlib.pyplot as plt # For plotting graphs
+import numpy as np # For numerical operations
+import os # For running a command in the terminal
+import pandas as pd # For data manipulation
+import platform # For getting the operating system name
+import random # For random number generation
+import re # For sanitizing filenames
+import seaborn as sns # For enhanced plotting
+import time # For measuring execution time
+from colorama import Style # For coloring the terminal
+from deap import base, creator, tools, algorithms # For the genetic algorithm
+from tqdm import tqdm # For progress bars
+from sklearn.ensemble import RandomForestClassifier # For the machine learning model
+from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix # For model evaluation
+from sklearn.model_selection import train_test_split # For splitting the dataset
+from sklearn.preprocessing import StandardScaler # For feature scaling
+
+# Macros:
+class BackgroundColors: # Colors for the terminal
+   CYAN = "\033[96m" # Cyan
+   GREEN = "\033[92m" # Green
+   YELLOW = "\033[93m" # Yellow
+   RED = "\033[91m" # Red
+   BOLD = "\033[1m" # Bold
+   UNDERLINE = "\033[4m" # Underline
+   CLEAR_TERMINAL = "\033[H\033[J" # Clear the terminal
+
+# Execution Constants:
+VERBOSE = False # Set to True to output verbose messages
+
+# Sound Constants:
+SOUND_COMMANDS = {"Darwin": "afplay", "Linux": "aplay", "Windows": "start"} # The commands to play a sound for each operating system
+SOUND_FILE = "./.assets/Sounds/NotificationSound.wav" # The path to the sound file
+
+# RUN_FUNCTIONS:
+RUN_FUNCTIONS = {
+   "Play Sound": True, # Set to True to play a sound when the program finishes
+}
+
+# Functions Definition
+
+def main():
+   """
+   Main function.
+
+   :param: None
+   :return: None
+   """
+
+   pass
+
+if __name__ == "__main__":
+   """
+   This is the standard boilerplate that calls the main() function.
+
+   :return: None
+   """
+
+   main() # Call the main function