Sahana Subramanian's repositories
Google-Play-Store-Analysis-and-App-Popularity-Prediction
Project on google play store app analysis (sizing and pricing strategy) | Bigram analysis of user reviews to discern patterns in user behavior and attributes of good/bad apps | Popularity prediction (install count) using random forest, decision trees and logistic regression
Engagement-and-Stock-Price-Analysis-of-CEOs-on-Twitter
Project on engagement and stock price analysis of CEOs on Twitter. Extracted the data from Twitter API and Yahoo Finance and implemented sentiment analyzer, topic modeling (LDA), stock price regression and engagement analysis to determine the factors that make a CEO influential
Track-Human-Footprint-in-Amazon-using-Deep-Learning
Project on multi-label classification of satellite images of Amazon rain forest using Deep Learning | Implemented deep CNN architectures along with haze removal techniques to achieve a F2 score of 0.9257 (top 20% of the Kaggle competition leaderboard)
Implicit-Recommendation-Engine-for-Meetup.com
Project on building implicit recommendation systems for Meetup | Built memory-based and model-based collaborative filtering (ALS and Logistic matrix factorization) recommendation engines using implicit feedback signals like RSVP count and timedelta | Data related to groups, events, members and RSVP extracted from Meetup API
Text-Analytics-Projects
A repository for projects related to text mining and natural language processing (NLP)
Analysis-of-NYC-Parking-Violations-using-PySpark
Project on analysis of ~10 million rows parking violations in NYC to explore the factors that might help prevent getting ticketed in NYC | The data was obtained from NYC Open Data and implemented in PySpark on DataBricks platform
Data-Science-Mini-Projects
A repository for various mini-projects as a part of my curriculum or personal interest | Includes data visualization, market segmentation, author attribution, portfolio modeling and association rule mining
3d-object-detection-for-autonomous-vehicles
Project on 3D Object Detection using Lyft's level5 dataset. Obtained mAP of 0.045 on the private leader board on kaggle and ranked in the top 20% among all teams participated in the competition.
cs231n.github.io
Public facing notes page
Data-Science-Coursera
Programming practice from Coursera
interpret_ml
Fit interpretable machine learning models. Explain blackbox machine learning.
STA380
STA 380: Predictive Modeling
tensorflow-lifetime-value
Predict customer lifetime value using AutoML Tables, or ML Engine with a TensorFlow neural network and the Lifetimes Python library.