a-b-testing

A/B testing on Udacity click stream data to reduce early course cancellations by using control and treatment groups

adtracking-fraud-detection

Analyzed dataset of over 200 million ad clicks and used machine learning techniques such as Logistic Regression, Random Forest, XGboost, SVM, and Recurrent Neural Network (RNN) to build model that could predict probability of an ad click being fraudulent.

analysis-and-visualization-of-road-traffic-accidents-in-usa

Analyzed 4 years of USA Accidents data with 3 million records, to help regular drivers and police to make calculated decisions

automated-data-pipeline-for-temperature-and-immigration-information

automatic-speech-recognition-

Built Automatic Speech Recognition (ASR) by combining a 2D Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and a Connectionist Temporal Classification (CTC) loss and evaluated the quality of model using Word Error Rate (WER).

aws-etl-data-pipeline-on-youtube-data-

ETL big data pipeline to manage, streamline, and perform analysis on structured and semi-structured data of more than 100,000 YouTube videos collected using YouTube API based on video categories and trending metrics

credit-cards-complaint-dashboard-

Developed and created a comprehensive credit card complaints dashboard using Tableau, analyzing data from 2015-2021 to identify the weekly trend, most common types of complaints, top companies with complaints, company response providing valuable insights for consumers and credit card companies.

credit-risk-modeling-and-scorecard-development

Developed data-driven credit risk model to predict probabilities of default (PD) and assigned credit scores to existing or potential borrowers using data includes information on over 450k consumer loans issued between 2007 and 2014 with 75 features.

customer-segmentation

RFM (Recency, Frequency, Monetary) model to determine which segments of customers should be targeted to enhance revenue for automobile bike company, by grouping customers into 11 segments based on previous purchase transactions.

customer-segmentation-and-market-basket-analysis-

Analyzed 3M+ grocery orders data from more 200k+ users to increase profitability by leveraging customer transaction behavior and purchasing history and discovered hidden association rules between products for better cross-selling and upselling.

data-analytics-customer-segmentation

In this project, a RFM model is implemented to relate to customers in each segment. Assessed the Data Quality, performed EDA using Python and created Dashboard using Tableau.

data-science

data225-covidvaccineeffect-project

Covid-19 vaccines' adverse effect analysis for Data 225-Database Management for data Analytics Final Project.

ecommerce-data-warehouse-in-hive-using-aws

Data warehouse for e-commerce application to perform Hive analytics on Sales and Customer Demographics.

fashion-smart-mirror-for-virtual-try-on-and-cloth-recommendation

Developed hybrid approach for image based virtual try-on by reconstructing 3D cloth model of the target cloth by matching silhouettes to body model, transferring the 3D cloth model to the estimated 3D target human model, generating target human skin parts and blend it to the rendered warped cloth along with human representations.

grocery-store-management-system-

Designed a user-friendly and interactive Power BI dashboard to gain insights into the sales data including product sales analysis, and customer segmentation of hardware company to drive informed business decisions and improve overall sales performance.

healthcare-provider-fraud-detection

Developed binary classification model using Provider data, beneficiary details, and inpatient/outpatient claims data to identify potentially fraudulent healthcare providers, helping to prevent financial loss and protect the healthcare system.

hybrid-recommendation-system-with-neural-network-

Hybrid recommendation system by combining Content-Based (Video Information), Collaborative Filtering (User Interaction), and Knowledge-Based (Personal information) for recommending videos and movies to the user.

image-and-video-segmentation-using-mask-r-cnn-

Performed Instance image segmentation using Mask Region-Based Convolutional Neural Network(Mask R-CNN) by generating bounding boxes, extracting segmented objects and detected target class.

information-retrieval-using-word2vec-based-vector-space-model

Performed information retrieval using Word2Vec based vector space model with 367k queries and 3.2M documents and performed pre-processing steps like converting text into lower case, expanding contractions, removing stop words, lemmatizing texts.

insurance-fraud-claims-detection-

Built machine learning model for enabling loss control units to achieve high coverage with low false positive rates to detect whether claim is genuine, or fraudulent based on data from an auto insurance company that has 40+ features and deployed using clouderizer.

marketing-campaign-analysis

Analyzed marketing data using Python and pandas to identify trends and patterns, performed exploratory analysis, generated summary statistics, visualized marketing reach by channel and language, and made informed decisions about marketing campaigns

multi-class-text-classification-with-transformer-models-

Classified textual data using BERT, RoBERTa and XLNET models by converting .csv datasets to .tsv format with HuggingFace library, and converting input examples into input features by tokenizing, truncating longer sequences, and padding long sequences.

ocr-in-python-with-opencv-tesseract-and-pytesseract

Extracted text from images and real time video by performing preprocessing steps like gray scaling, noise removal, thresholding, dilation, erosion, rescaling, deskewing, canny edge detection, template matching and added bounding boxes for all the characters.

ocr-with-tesseract

A comprehensive tutorial for OCR in python using Tesseract-OCR and OpenCV

predictive-price-modeling-

Predictive Price Modelling and Alternate Search Rankings for Airbnb listings

python-basics

real-time-twitter-sentiment-analysis-for-brand-improvement-and-topic-tracking

Implemented data pipelines using streaming Twitter API, Tweepy, OAuth, SQL, Python to process 5M+ daily tweets. Automated sentiment analysis using RE, TextBlob, and NLTK based on tokenization, lemmatization, and stemming and created an interactive dashboard using Plotly and Tableau.

retail-sales-dashboard-using-plotly-and-dash-

Developed a retail sales dashboard using Plotly and Dash to visualize and analyze sales data, allowing for interactive exploration of sales trends by month and day.

sale-performance-dashboard-using-looker-

Developed a sales team performance dashboard using Looker Studio and Google Sheets, analyzing data from 2010 to 2015 to track key metrics and KPIs and provide actionable insights into sales trends and performance over a five-year period.