Interpreting direct sales’ demand forecasts using SHAP values

Paper aims Several concerns regarding the lack of interpretability machine learning models obstruct implementation projects as part demand forecasting process. This paper presents a methodology to support introduction into process traditional direct sales company by providing explanations for otherwise obscure results. We also suggest incorporating human knowledge inside pipeline an essential capturing business logic and integrating existing processes. Originality Using explainable methods on real-life data demonstrates that techniques are functional beyond academy can be introduced everyday companies' production. Research method The project used real-world from followed collect, preprocess, select train model, conclude with explanation model results through SHAP Main findings provided insights contribution features forecast. analyzed individual predictions understand behavior different variables, proving helpful when interpreting complex models. Implications theory practice study contributes discussion about adopting new technology implementing forecasting. presented in this implement similar interested companies.