Cindysetey/measles2-uptake-predictors

Capstone project applying machine learning to identify key predictors of Measles 2 vaccine uptake among under-five children in Kenya. Focused on improving immunization strategies and public health interventions. # Detecting Key Predictors of Measles 2 Uptake Among Under-Five Children ## Project Overview This repository contains my capstone project focused on detecting the **key predictors of Measles 2 vaccine uptake** among children under five years in Kenya. The notebook applies **data cleaning, exploratory data analysis (EDA), and machine learning models** to identify factors influencing immunization coverage. The project’s insights aim to support **public health decision-making** and improve vaccine uptake strategies. --- ## Objectives - Clean and preprocess the dataset on Measles 2 vaccination. - Perform **EDA** to understand patterns and distributions. - Apply **machine learning models** to predict vaccine uptake. - Identify the **most important predictors** influencing uptake. - Generate insights for **public health interventions**. --- ## Methods and Tools - **Programming Language**: Python - **Libraries**: - `pandas`, `numpy` → data manipulation - `matplotlib`, `seaborn` → visualization - `scikit-learn` → ML models & evaluation - `imblearn` → class balancing (SMOTE, undersampling) - **Models Implemented**: - Logistic Regression - Decision Tree - Random Forest - **Evaluation Metrics**: - Accuracy - Classification Report (Precision, Recall, F1) - Confusion Matrix - ROC-AUC --- ## Dataset - **Source**: Kaggle - **Population**: Children under five years in Kenya. - **Features**: Socio-economic, demographic, and health access variables. - **Target Variable**: - Binary outcome → Measles 2 uptake (`1 = Uptake`, `0 = No uptake`). --- ## Workflow 1. **Load dataset** and preview records. 2. **Data cleaning**: handle missing values and drop unnecessary columns. 3. **Feature engineering**: - Create binary target variable. - Encode categorical variables. 4. **EDA & visualization**: distributions, correlations, bar plots. 5. **Modeling**: train Logistic Regression, Decision Tree, and Random Forest. 6. **Evaluation**: compare models using classification report and conf …