Competition · Python / Software Engineering
Demand forecasting
An anonymised ML competition case: data preparation and forecast evaluation.
- Python
- scikit-learn
- Optuna
An anonymised competition case. Competition datasets and materials are not published.
Project and contribution
Worked on feature preparation, model comparison and parameter search for demand forecasting. Evaluated approaches on time-based splits and analysed the limits of the results.
Task
A demand-forecasting task in an ML competition. The goal was to prepare data, compare approaches and evaluate how models perform on later time periods.
My work
- Prepared data and modelling features.
- Compared models and searched parameters.
- Used time-based splits to evaluate forecasts.
- Analysed limitations of the approaches and evaluation results.
Technical approach
Used Python, scikit-learn and Optuna. This description covers only the general approach. Code, data, plots, numerical results and organiser details are not published.