Provide machine-learning model-based demand forecasting for Chinese pharmaceutical companies

This website will provide an overview of what we accomplished in this Capstone project.

Read Our Full Report


 

Problem Statement

Our Capstone project was collaborated with a Chinese Pharmaceutical company. 

Their business model were outdated and in order to help them better track the demand in the market, we aimed to provide a machine-learning model based smart demand forecasting. 

Our results was made based on their sales & purchase database from 2013 to 2019.

Methods

Here are the overall stages in our Capstone project.

  • Data Validation
  • Feature Engineering
  • Time Series Analysis (ARIMA)
  • Machine Learning Approach
  • Future Improvement
  • Conclusion

 

Summary

In the era of rapid competition, being able to respond fast to market changes, maintain stable partnerships and customize marketing strategies are the primary conditions for companies to be one step ahead of the competition participating in the market games.

The objective of this project is to perform descriptive and prescriptive analyses of a Chinese pharmaceutical company’s dataset in order to help them with demand forecasting and customer pattern recognition utilizing Statistical Analytical models as well as Machine Learning models. 

The time series analysis results have shown that the unit price of the top-selling products is stationary and contains strong seasonality. By implementing the entire machine learning pipeline, we made the sales prediction of the top fifteen selling products within 95% of the confidence interval for our clients to help them better understand their product.