Situation
Our client, a European cycling and motor parts retailer, wanted to be able to predict the faults of different cars. The question from our client was: “Given where the vehicle is based, what we know about individual vehicle makes, models and ages, beyond perishables, what parts are likely needing to be replaced or services applied for this vehicle or a vehicle like it?”
Approach
We built an analytics platform where the following hypotheses were used to predict future faults in cars: cars belonging to the same family share common componentry and inherit common problems. Errors in data may skew results, clustering ‘similar looking’ vehicles allows for a robust approach. A vehicle’s location affects issues related to corrosion caused by salt and moisture in the air. The data of the hypotheses were the basis of the Common Vehicle Issues solution.
Results
Our client now knows in advance what issues cars coming into the garage have and will likely have in the future. This has resulted in an increased turnover and more satisfied customers, who could have issues fixed in advance, resulting in fewer breakdowns.
