SharmaineMangombe/Childline-Kenya-Call-Volume-Prediction-Challenge

# Childline Kenya Call Volume Prediction Challenge ## Overview This project is part of the Zindi competition to predict the **call volumes at Childline Kenya**, a helpline for children in need. The goal was to forecast call volume to help the organization **allocate resources efficiently during peak hours** and improve service delivery. ## Problem Statement Childline Kenya experiences varying call volumes daily. Accurate predictions allow the organization to: - Optimize staffing - Reduce response times - Ensure that children in need receive timely support The challenge was to build a time series model to forecast call volumes based on historical data. ## Tools & Technologies - **Python:** Pandas, NumPy for data processing - **Time Series Analysis:** ARIMA, Prophet, and other forecasting methods - **Data Visualization:** Matplotlib, Statsmodels - **Data Cleaning & Preprocessing:** Handling missing values, outliers, and formatting ## Methodology 1. **Data Collection:** Imported and explored historical call data from Zindi dataset 2. **Data Cleaning:** ensured correct date formatting and Transformed the date column 3. **Exploratory Data Analysis (EDA):** Visualized trends, seasonal patterns, and peaks in call volume 4. **Model Development:** - Tested multiple time series models (ARIMA, Prophet) - Tuned hyperparameters for best prediction accuracy 5. **Model Evaluation:** - Measured performance using Mean Squared Error (RMSE) 6. **Prediction & Insights:** Generated forecasts for call volume, highlighting peak periods for resource planning ## Results - Successfully predicted daily call volumes with acceptable error margins - Identified **peak call periods**, allowing for **better staffing and resource allocation** - Insights from this project could help reduce response time for children in need