Goal

▪ To predict when the next equipment repair based on past repairs.
▪ Then proactively approach customers to maintain and fix the equipment before equipment repair happen.

ML Model

A tree-based regression model with confidence intervals (upper and lower bounds) was implemented to predict the next equipment repair.

Project Outcomes

▪ Identified top pieces of equipment and the top customers based on the quality and the quantity of data available.
▪ Strategically identified and built methods to filter out bad data. (The percentage of bad data was too high)
▪ Built an Alteryx workflow to clean the data and predict when the equipment will next fail.
▪ Provided an upper bound and a lower bound of the next equipment repair.
▪ Identified scopes for data improvement.

Results

• Developed a tree-based regression model with confidence intervals (upper and lower bounds) for the prediction.
• The model predicts the next equipment repair (+/- 30 days) with 75% accuracy.