The Challenging Scenario
The critical key to future success – a keen eye to spot and sort public companies by their real performances!!!Share To:
An investment bank needed an agile data-model driven technology to score different public companies based on a proprietary algorithm. The goal was to use historical stock prices and published financial metrics to produce a score that described how likely a particular company was to increase its shared price as well as how environmental any particular company was.
PySpark driven custom data model for a mammoth historical data store of 3000 companies!!!Share To:
It wasn’t taking long for Transpire team of core-competent financial technocrats to quickly mobilize and swing into action. We worked with a financial data provider to ingest historical data via an API. The data was ingested and persisted in a S3 bucket. Once all historical data for all the 3000 large public companies listed in different stock markets in the US were saved, our team set up a PySpark job that read each file, transformed them into a data model customed developed by our data architect and saved them back into another S3 bucket.
Resolution in detail
The right technology does it all – extract concrete, quantifiable insights out of 3+ Terabytes datastore; build powerful, flexible custom ML models!!!Share To:
Athena was our data access layer.
Our experts developed custom analytics query to aggregate and extract additional insights out of the over 3 Terabytes of data stored.
Once the dimensionality of the data was reduced it was then saved in another Athena table in order to run custom ML models using EMR clusters, python, pythorch and PySpark.
Ground-breaking business impacts
The new makeover with stellar results– isn’t it a delight to cut down huge operating cost and incur only runtime expenses?Share To:
The ML models would be re-run once a week after the markets closed on Fridays. Having this setup allowed our client to reduce operating expenses as the cost was incurred only at runtime.