software engineer - data scientist
TRANSCRIPT
Experience
Dheeraj Kura – Data Scientist, R&D Engineering, Megasoft Limited
Contact – 8886560505, e-mail: [email protected] , [email protected]
Education:MBA – Osmania University (Finance)Graduation -B.Com (Comp)Certification: Certificate Program in Big Data Analytics and Optimization from INSOFE
Projects Delivered so far
Failures Prediction
- Telecom, Calls Drops is Frequent in the Processing due to heavy Load on the Tower which carries the Spectrum to process the calls. Prediction of the Possible Failure Reason and alerting to share the load to nearest Tower
Benefits: Load Balancing, Error Prevention, Enhanced Quality of Service
Algorithms Used –
Regression – To find the Possible Reason for Failure
Time Series – Data was regressed with time to find the patterns of the System performance and correlating the same with the Unforeseen Output which has the prediction accuracy of 82%
Customer Lifetime Value
- For Every Business it’s the Key factor to knowhow valuable the customer is , using the Predictive Analytics same was estimated for the Future
Benefits: To enhance Effective Marketing Strategy and also to implement plans to attract New Customers
Benefits: Improved plans to retain the Customers/Increase loyalty and also to implement plans to attract New Customers
Algorithms Used
Time Series: Data was regressed with time to find the patterns of the Customer Usage of the Service for Calls, SMS & Data
Prior Experience
Cognizant Technology Solutions, Hyderabad (Sept 2013 – Sept 2015)Senior Process Executive - Dispute Management –client: UBSTeamed up in Regulatory On boarding which is dedicated for Dodd Frank & EMIR regulatory changes
Bank of America Continuum Solutions (Aug 2010 –Sept 2013)Senior Team Member – Dispute Management
Worked with Collateral Team aimed to Maintain Required Collateral to cover Market Risk
Project: Prior authentication prediction for prescribed Medication (Health Care, Insurance)
Our goal is to develop an engine, for doctors that tell whether prescribed medication will require a prior authorization or not. This helps the doctors to prescribe those medicines that do not require Prior Authorization in turn increases the probability of purchasing the medication. This engine should be based on advanced machine learning technologies that looks at the past transactional data and predict with high degree of confidence whether a particular drug will require prior authorization for a particular patient or not.
Data Description:
Data Contains Drugs and its Sub Types which are having Multiple Levels
Techniques used: Logistic Regression, Clustering, Bayesian techniques, Decision Trees Finally Ensembling Technique is used to fit a best model
- Software environment: R
Synopsis: In the Traditional approach, accuracy was achieved was 92% through this engine accuracy improved to 95%