Problem Statement
- Ingest large volume (multi-TB/week) of mobile and broadband usage data from a third-party service provider.
- Perform market research on full datasets (instead of samples).
Solution Overview
- Foundational Data Lake setup using PaaS services on MS Azure
- Data science modelling was deployed using python to predict the call drops on Azure data lake.
- Ingestion of multi-TB data / week from 3rd party source (P3)
- 3 billion data points from 150,000 mobile devices analysed.
- Rapid on-boarding of template driven approach, Data Standardization, Cleansing, Validation & Auto-profiling framework using our Accelerator.
- Governance UI for Metadata Search, Lineage analysis, manage objects, analyze profiling statistics and perform operational reporting.
Outcomes
- Data analyzed from the research helped client in policy making
- Mobile and broadband service assessment across multiple dimensions – service providers, technology, geo, data/voice
- Call drop prediction capabilities provided via MLXpress
