2024 Flex-Metrics Team

The Challenge

MSBA-BA graduate students Tyler Scott, Esther Wang, Kayla Jones, Isaac Moore, and Brendan Atkisson took on the challenge of predicting machine failures for Flex-Metrics, a company that provides real-time tracking of manufacturing outputs. Their goal was to develop a predictive analytics method that could be integrated into Flex-Metrics’ software to help manufacturers prevent costly machine downtime. This would be achieved through a machine learning model and a Power BI dashboard that would deliver actionable insights to Flex-Metrics’ diverse client base.

The Solution

The team developed a machine learning model using a random forest algorithm, chosen for its ability to accurately predict machine failures while minimizing false positives. This balance was crucial to ensure user engagement with the tool. They transformed and analyzed data provided by Flex-Metrics, focusing on 26 unique metrics over five shifts of data to create a robust predictive model. The model was integrated into Flex-Metrics’ software, with predictions refreshed before each shift via a Python script connected to a SQL database. The resulting Power BI dashboard provides daily risk assessments, showing machine risk levels and top predictors of potential failures. This solution not only increased product value and user engagement for Flex-Metrics but also introduced the company to the world of predictive analytics, offering new opportunities for future product development.