Sumit Asthana's Portfolio

Title Description Site Url Image Skills
Order Fallout Analysis for Global Cable Television Company
• Parsed unstructured webAPI data using python to convert them into structured format. • Using structured data, order derailment was analysed. • The corresponding errors were retrospectively... Read More
• Parsed unstructured webAPI data using python to convert them into structured format. • Using structured data, order derailment was analysed. • The corresponding errors were retrospectively analysed and categorized based on Customer Touchpoints, Contracts and Error Categories and visualized in D3.js. • Analysed Customer Experience Feedback Using Text Mining and extracted key reasons for order fallout. Read Less
Python
Early Delinquency Scoring
• Many of the loans were turning delinquent early during term of the loan. • Loans were identified with high chance of being delinquent and scored. • Identified loans were allocated resources ... Read More
• Many of the loans were turning delinquent early during term of the loan. • Loans were identified with high chance of being delinquent and scored. • Identified loans were allocated resources in order to strategize the retention. • Exploratory data analysis i.e. data distribution, IV and WoE values, correlation and ?2 values. Dummy value creation, binning and new feature creation. • Model fitting - Random Forest, Logistic Regression. Read Less
R
Expansion Analysis
• Analyzed App data and scrubbed it make analytics ready. • Clustered consumers based on device expansion and built a classifier to target customers having high propensity to expand.
• Analyzed App data and scrubbed it make analytics ready. • Clustered consumers based on device expansion and built a classifier to target customers having high propensity to expand. Read Less
R, Python
Paid-In-Full/Recapture Probabilities
• Mortgage bank originates and services home loans from primary and secondary market. • Scoring of customers which has propensity of paying the remaining amount on loan. • Campaign model will ... Read More
• Mortgage bank originates and services home loans from primary and secondary market. • Scoring of customers which has propensity of paying the remaining amount on loan. • Campaign model will be based on scores to recapture the customers. • Exploratory data analysis i.e. data distribution, IV and WoE values, correlation and ?2 values. Dummy value creation, binning and new feature creation. • Model fitting - Random Forest Read Less
R, Python
Social Media Mining
• Client is the leader in the lighting control industry and offers a wide selection of energy saving dimmers and lighting control solutions. • Analysed the corpus of Amazon reviews and produced a... Read More
• Client is the leader in the lighting control industry and offers a wide selection of energy saving dimmers and lighting control solutions. • Analysed the corpus of Amazon reviews and produced actionable insights. Used NLP techniques for topic mining and sentiment analysis. Classified reviews based on user ratings for each category which were generated during themes generation exercise. • Data Sources: Reviews from Amazon about the product. • Data Processing, Data Review, Modelling and Business Insights. • Model Fitting – Topic modelling using LDA & NMF, Supervised sentiment analysis. Read Less
Python