How rosiemin/contraceptive_product_forecast works
This repository is an analytical forecasting project for monthly contraceptive stock distribution at Côte d’Ivoire health sites. Its main workflow loads and joins inventory and reference data, filters site-product series, then evaluates univariate forecasts and a forecast-distance regression approach for predictions. The README also describes exploratory analysis and verification metrics. The graph includes these documented stages, but source-level relationships are limited to the excerpts and README; the exploratory notebooks and some outputs were not inspected.
Main entry points: clean_dta.py Load and clean · forecast_cv.py Forecast-distance models · univariate.py Univariate models · site_product_univariate.py Site-product forecasts
Components 12
Data preparation
- Load and clean
src/clean_dta.py - Series filtering
src/clean_dta.py
Forecasting
- Univariate models
src/univariate.py - Site-product forecasts
src/site_product_univariate.py - Lag features
src/forecast_cv.py - Forecast-distance models
src/forecast_cv.py
Analysis and validation
- Exploratory analysis
notebooks/EDA.ipynb - Image analysis
src/image_code.py - Verification metrics
src/univariate.py - Forecast results
results/submission_predictions.csv
Other components
- Analyst
- Input datasets
Connections 12
- Input datasets → Load and clean: loads and joins
src/clean_dta.py - Load and clean → Series filtering: filters series
src/clean_dta.py - Analyst → Exploratory analysis: explores data
README.md - Load and clean → Image analysis: prepares data
src/image_code.py - Load and clean → Site-product forecasts: provides data
src/site_product_univariate.py - Site-product forecasts → Univariate models: calls models
src/site_product_univariate.py - Univariate models → Verification metrics: computes metrics
src/univariate.py - Load and clean → Lag features: provides data
src/forecast_cv.py - Lag features → Forecast-distance models: creates features
src/forecast_cv.py - Forecast-distance models → Univariate models: uses metrics
src/forecast_cv.py - Forecast-distance models → Forecast results: writes predictions
src/forecast_cv.py - Verification metrics → Forecast results: records scores