I come from a family which have had history of having commerce as majors, thus the institutional love for mathematics comes inherently. But there was also other part of growing up where I used to invest most of my time gaming on my PC specifically NFS, Age of Empires, etc. I always had a deep desire for knowing how these games are made or how to make my own program which works on click of a button. As the time progressed, I kept on developing my knowledge in Math from basic fractions leading to integrations and in programming starting from language as LOGO, BASIC then moving to JAVA, COBOL.
This passion carried me to get a job at TCS where I learned about a new field which would totally connect with me, Data science. Here the coding part would be the front face for my analysis of real-world problem whose solution would be achieved through mathematical approach. But to learn and amplify my knowledge I pursued my masters at UIC. Here there were more resources available to me than I could have ever imagined as a kid. I learnt new languages, algorithm, compilers, higher mathematics, all with the same enthusiasm that drove me since my childhood.
The zeal to grow keeps getting better and is leading to me where it always led learn more and more.
-
Experience
Dedicated analytics professional with over 4 years of experience in Financial Services Analytics.
I have worked in various stages of project life cycle from analyzing the business requirements and functional models to building, integrating and data reporting using SQL, and Excel.
Tools: SAS (Base), R (dplyr, caret, ggplot2, SVM), Python (numpy, pandas, NLTK, sci-kit-learn, pyspark, pytorch), SQL, Excel, AWS EC2
Machine learning techniques: Hypothesis Testing, Data Mining, Linear Regression, Logistic Regression, GLM, Decision Tree, Random Forest, Data Visualization, XG Boosting, Clustering, Naïve Bayes Classification, Recommendation Engines, NLP
Deep Learning techniques: Feed forward Neural Network, Convolutional Neural Network, Epsilon Greedy, Upper Confidence Bound, Thompson Sampling, Reinforcement Learning.
-
Projects
Fake Tweet Prediction| Python
• Applied Regression and classification analysis using stemming, lemmatization and bag of words to predict whether a tweet is fake or not based on data collected from database of genuine tweets.
• Implemented Random Forest, SVM, Naïve Bayes and GBM learning models using Python.
Loan defaults predictions for Lending Club| R Studio
• Developed a predictive model to predict which loans are at risk of fault using Random Forest, GBM and GLM.
• Evaluated performance of the model using metrics like Confusion Matrix, AUC, ROC, Gini Coefficient.
Best arm prediction in a Multi-Arm Bandit |Python
• Applied the concept of reinforcement learning by learning over the performance of a multi-armed bandit problem to predict which arm will be most successful.
• Used the approaches like epsilon-greedy, upper confidence bound, Thompson sampling techniques to determine best arm.
New York Taxi Fare Prediction| Python
• Predicted the taxi fare of New York City taxi services based on pick up, drop location using regression concepts.
• The model performance was evaluated using metrics like Linear regression, Gradient boosting, Random forest in python environment.
Cancer cells size determination | Python and AWS EC2 instance
• Deployed a model which predicts whether the cancel cells are benign or malignant based on Amazon Web Services.
• The model performance was evaluated using metrics like SVM and was deployed on AWS Ubuntu EC2 instance using flask.