SITUATION​
Our client faced a significant challenge in accurately predicting store sell-through rates, which was crucial for effective inventory management and optimizing product availability. They recognized that machine learning could offer the potential to predict future outcomes based on historical data, but they needed specialized skills and tools to utilize these technologies effectively. ​

The goal was to create a machine learning model that could predict when a store would sell out of a specific Stock Keeping Unit (SKU), allowing them to adjust inventory quantities accordingly. The client engaged Everforth Apex for our advanced expertise in data and machine learning solutions.​

SOLUTION​
Everforth Apex proposed developing several machine learning models with slightly different characteristics using Google Cloud Platform (GCP) tools and evaluating them with real historical data. The team employed tools and codes such as Python (with Google Colab), Vertex AI, and forecasting models such as ARIMA and Prophet (used by Facebook). The approach involved analyzing various options and testing different methods to achieve higher accuracy in the model’s predictions. The team tested various methods until they found the tech stack that would work best with the client's existing data models. ​

The project was led by a team of data analysts who worked closely with the client's engineering team to ensure seamless integration and collaboration.​

 

RESULT​
The developed models have demonstrated high accuracy in predicting store sell-through rates, providing valuable insights into workflow development and tool evaluation. ​

Although the project is ongoing, the initial results are promising, and the client has decided to continue working on this initiative to achieve even better outcomes, leading to logistical savings and better stock movement. This exploration lays the foundation for future research into more robust and effective solutions.