Breast Cancer Prediction Using Machine Learning

Overview

This project aims to predict breast cancer diagnoses (Malignant or Benign) using machine learning models based on tumor characteristics. Data preprocessing, model training, and AutoML techniques were employed to achieve high accuracy and optimize performance.

Dataset

The dataset contains tumor features such as radius, texture, and perimeter, with the target variable being the diagnosis (Malignant or Benign). The dataset can be accessed here:https://www.kaggle.com/uciml/breast-cancer-wisconsin-data

Workflow

1. Data Preprocessing:

   • Cleaned data by dropping missing values and encoding diagnosis labels (M → 1, B → 0).
   • Standardized features using StandardScaler.

2. Model Training: Trained multiple models, including:

   • Ridge Classifier
   • AdaBoost Classifier
   • Extra Trees Classifier
   • Decision Tree Classifier
   • Random Forest Classifier

3. AutoML with PyCaret:

   • Compared models using PyCaret's setup() and compare_models() functions.
   • Tuned the best-performing model (Random Forest) to maximize accuracy.
   • Visualized results with confusion matrices and feature importance plots.

4. Model Evaluation: The Random Forest model achieved the best performance with high precision, recall, and F1-score. The model was saved for future predictions.

Technologies Used

• Python, Pandas, NumPy, Scikit-learn, PyCaret
• Visualization: Matplotlib, Seaborn

Results

The project successfully predicted breast cancer diagnoses with high accuracy, showcasing the power of machine learning in healthcare applications.

---

For more information, contact me at:

• Email: [email protected]
• LinkedIn: https://www.linkedin.com/in/himani-s1001/