Ukayria/cholera-outbreak-predictor

This model predicts Cholera outbreaks by state in Nigeria. # Cholera Outbreak Predictor 🦠📊 This AI-enabled project was created during my 3MTT learning journey. It attempts to predict potential cholera outbreaks in Nigeria by leveraging machine learning with historical weekly data. ## Why This Project? Cholera remains endemic in Nigeria. During my learning journey, I created this project to understand the role of AI in supporting public health initiatives. I developed a classifier based on data pulled from the Nigerian Centre for Disease Control (NCDC) Weekly Epidemiological Reports (2021–2025) that marks weeks as **“Outbreak”** or **“No Outbreak”**. ## Project Structure - `Cholera_outbreak_predictor.ipynb`: Main notebook containing the complete workflow for data analysis, including data cleaning, exploratory analysis, model creation, and evaluation. - `Cholera_Report_Cleaned_new.csv`: Preprocessed dataset used for model building and analysis, compiled from NCDC Weekly Epidemiological Reports. - `README.md`: Project overview, objectives, methodologies, and workflow. - `.pkl` files: Trained logistic regression model(s) for prediction. ## What the Project Covers - **Data Collection**: Weekly cholera reports from NCDC including suspected and confirmed cases, deaths, and case fatality rates. - **Data Cleaning & Exploration**: Checked for duplicates and missing values; explored patterns using data visualization. - **Feature Engineering**: Created new features from existing relevant data to improve model performance. - **Model Training**: Tested multiple classification models; chose Logistic Regression for its reasonable accuracy. - **Model Evaluation**: - *No Outbreak (Class 0)*: Recall = 1.00, Precision = 0.92 - *Outbreak (Class 1)*: Precision = 1.00, Recall = 0.50 - Threshold adjusted to improve outbreak detection. ## Insights The model performs well in detecting weeks with no outbreaks. With threshold tuning, its ability to detect potential outbreaks improves significantly. ## Tools & Technologies - **Platform**: Go …