Data science Portfolio

Repository containing portfolio of data science projects completed by me for academic, self learning, and hobby purposes. Presented in the form of iPython Notebooks, and R markdown files

Graduating Project: Sentiment Classification and change detection of online restaurant reviews using Transfer Learning and Difference in Difference Modelling.

Internship: End of the year Artificial Intelligence internship at IFSTTAR, Nantes in Sensory data pattern recognition and wheel classification.

Machine learning Projects

1. Product Demand forecasting and Customer segmentation: Thus, it is in Retailer’s interest to forecast the demand of the products based on the trends in past transactions and at the same time obtain customer segments by considering their demographics and past purchases for target marketing.
2. Bank Transaction classification: Utilizing word embeddings and SVM to classify categories using Bank Transaction descriptors.
3. Content Recommendation systems: This project implements Sequestial experimentation design to build robust content recommendation models. In this project, I also evaluated different libraries to understand their strenghts and provide a comparision of the libraries from an end-to-end model development perspective.
4. Bankrupt or Not : Predicting given Firm is going to Bankrupt or not using Financial Data. The given data contains financial ratios of various firms and the target attribute tells whether the firm has got bankrupt in the subsequent years or not.
5. Customer Revenue Prediction: The CRM team of retail store wants to understand the influence of given attributes on the revenue generated by the customers and to come up with a predictive model which helps to know the revenue (Amount) of a given customer using the model.
6. Heart Disease Classification: A Healthcare organization together with a couple of government hospitals in a city has collected information about the vitals that would reveal if the person might have a coronary heart disease in the next ten years or not. Given the data, we need to develop appropriate models to identify/predict if the person likely to have heart disease or not.

Computer Vision Project

1. Barcode detection: Barcode detection using YOLO approach and DarkNet. Detecting Barcodes on Text-books for information assimilation and retrieval.
2. Indian Cuisine detection application: Kivy application which can detect 20 Indian dishes in Images.

Automation and bots

1. Telegram Cypto Sentiment Bot: Telegram bot using Naasai (An open-source contribution demonstrating Naasai's capabilities to get crypto sentiment on Newsapi articles)

Social Media Intelligence

1. Sentiment and Network Analysis of Different groups on Autonomous Vehicles: The aim of this project is to find people’s, governments’ and companies’ opinion regarding autonomous vehicles by performing Sentiment and Network Analysis using Twitter dataset.

Kaggle Kernals

1. Household Load Forecasting-Stats+LSTM+Prophet: Load forecasting using Holt-Winters, ARIMA, Long Short-Term Memory Networks(LSTM), and Facebook's Prophet.
2. Comprehensive Tutorial on Churn Modelling: Churn Modelling using classification algorithms and producing rules for targetted Marketing.

Hackathons

1. Wind Power forecasting for the day-ahead energy market by Compagnie Nationale du Rhône
2. Zindi Economic Well-being prediction challenge

Data Visualization and Analytics

1. Dognition Company Smart Recommendations using Tableau (click on the gif for the project presentation)