/FeatureSelection-GeneticAlgorithm

Dive into feature selection and classification with this Python repository, utilizing a genetic algorithm and various classifiers on a breast cancer dataset. Achieving high accuracy levels, SVM and ANN stand out with 97.37%. Ideal for machine learning enthusiasts and those interested in cancer diagnostics.

Primary LanguagePython

Feature Selection Genetic Algorithm and Classification

This repository contains Python code for feature selection using a genetic algorithm and various classification algorithms applied to the well-known WDBC dataset (Wisconsin Diagnosis Breast Cancer dataset). The dataset is loaded from a CSV file using pandas, and preprocessing steps include handling missing values and label encoding.

Feature Selection with Genetic Algorithm

The genetic algorithm is implemented using the DEAP library to evolve a population of binary strings representing selected features.

Classification Algorithms

The selected features are used to train SVM, Random Forest, Decision Tree, and Artificial Neural Network (ANN) classifiers. Additionally, a Multi-Layer Perceptron (MLP) classifier is included for comparison.

Evaluation Metrics

The accuracy, precision, recall, and F1-score are calculated for each classifier. Confusion matrices are generated for the ANN classifier, and precision-recall metrics are computed for the MLP classifier.

Dependencies

The code utilizes popular Python libraries such as NumPy, scikit-learn, pandas, Matplotlib, Seaborn, DEAP, Keras, and others. The data can be loaded either locally or from Google Colab.

How to Use

  1. Environment Setup

    • Make sure you have access to Google Colab.
    • Upload the dataset (data.csv) to your Google Colab environment.

  2. Install Dependencies

    • Open a new Colab notebook.
    • Install the required libraries in requirements.txt file

  3. Run the Code

    • Copy the provided code into separate cell (or cells) in your Colab notebook.

  4. Adjust Parameters (Optional)

    • You can customize parameters such as population size, generations, crossover probability, and mutation probability to suit your preferences.

  5. Review Results

    • After running the code, the results, including classifier accuracies and evaluation metrics, will be printed in the Colab output.

  6. Explore and Experiment

    • Feel free to experiment with different datasets, tweak the genetic algorithm parameters, or try other classifiers to see how the system behaves.

Takes into account that Google Colab provides a pre-installed environment for many popular libraries, and you only need to install additional dependencies listed in the requirements.txt file.