BIG DATA PLATFORM
Setting up a Reliable and Powerful Platform
Platzhalter
Icon 1
Platzhalter
Icon 2
Platzhalter
Icon 3
INITIAL SITUATION
Our client has been collecting a huge amount of manufacturing and maintenance data and was struggling to make sense of it. The vision have been to build a predictive maintenance platform to save costs through reducing expensive repairs​ by catching failing components before they break.
SOLUTION
We have been developing a predictive model based on a time- dependent training set. The ML model is able to predict possible issues for a specific machine based on many different features (when was it produced, when was it sold, components, etc. ….). For each prediction confidence interval on prediction are also provided to increase model interpretability.
BUSINESS VALUE
Our customer has been able to better visualize and analyze maintenance issues as well as contact several customer to replace specific components while still in warranty period. The Operation teams was able to recognize patterns reducing significantly the high costs for repairement.
FRAMEWORK & TOOLS
The entire pipeline has been developed using big data technologies (Spark, Hadoop, Python) with a final visualization dashboard in Tableau. Finally, the project has been delivered to the operation team after the production rollout.