Release Notes for Breast Cancer Prediction Model - Version 1.0.0

Release Notes for Breast Cancer Prediction Model - Version 1.0.0

Date: 28th August 22, 2023

Overview:

In this release, we introduce the Breast Cancer Prediction Model, a machine learning project aimed at predicting breast cancer cases using data from the KNUST hospital survey. This model has been meticulously crafted after rigorous data wrangling, visualization, and hyperparameter tuning, resulting in an accuracy of 90.57%.

Highlights:

1. Data Acquisition: Sourced data from KNUST hospital breast cancer survey.
2. Data Processing: Comprehensive data cleaning and wrangling to ensure data integrity.
3. Visualization: Employed various visualization techniques for insightful data exploration.
4. Feature Selection: Used the Chi-Square approach to derive the most impactful features.
5. Modeling: Developed a Random Forest model with extensive hyperparameter tuning.
6. Evaluation: Achieved an accuracy of 90.57%, with precision, recall, and F1-score values indicating a robust performance.

New Features:

• SMOTE Upsampling: Improved model's performance on imbalanced datasets by integrating SMOTE upsampling technique.
• Hyperparameter Tuning: Employed grid search techniques for optimal parameter discovery, refining model performance.
• Bias-Variance Trade-off Analysis: Analyzed the trade-off to ensure a well-balanced model, minimizing overfitting.

Dependencies:

• Introduced tensorflow==2.5.0 for potential deep learning explorations.
• Relied heavily on scikit-learn==0.24.2 for modeling and data preprocessing.