/medical-sales-forecasting

Machine Learning | A Data-Driven Sales Forecasting Model For Medical Products In Africa | Random Forest Classifier & LASSO

Primary LanguageJupyter Notebook

Machine Learning | A Data-Driven Sales Forecasting Model for Medical Products in Africa

In this data-driven operations management project, the sales volumes of pharmaceuticals have been forecasted over time. The database contained 22,385,920 sold units of 9,128 products with 2,022 strengths from 11,240 brands which have been recorded by an African pharmacy network of 300 African pharmacies between January 2017 and June 2019. Due to very different sales volumes over time and across stock-keeping units, sales quantities are highly uncertain. A random forest classifier in combination with a multivariate linear regression with a LASSO regularization was applied for achieving the sales forecasts.

The project includes:

  1. Data analysis
  2. Data cleaning, data preprocessing and feature engineering
  3. Model selection and model evaluation
  4. Development of a GUI for visualizing the sales forecasts

Please contact me for more information!