Intelligent Device Data Monitoring

September 6, 2020

Manufacturing conglomerate

Problem Statement

• Over 10K monitoring devices generate data across site locations and different client, which is generating device log data minute by minute. All this is consolidated daily and must be standardized and pulled in for correlation with a very tight TAT
• The existing process is a manual process and utilizes a very primitive Machine Learning conceptual approach. There is a very low accuracy and prone to many manual errors
• A partial system for intelligent machine learning, standardization and scale up for all the devices and their data was built

Analytics Led Approach

• Using apriori matching algorithm, which utilizes intelligent machine learning concepts for document matching and classification
• Productizing this approach for scalability and work on multiple files, all simultaneously
• High accuracy of the algorithm, to help match multiple lines of text using fuzzy matching and document frequency counts

Critical Success Factors

• High accuracy of match and standardization, upto 100% in majority of cases
• Quick TAT with scalable algo running 500K lines in less than 45 minutes
• Ability to scale up with codes written in SAS and using intelligent priority based matching approach