samratsengupta Goto Github PK

amazon-apparel-recommendation

build content based recommender from derived feature set of data of popular retail giant using idf weighted w2v based semantic similarity for text metadata and imagenet trained VGG-16 for image recommendation

apparel-search-bot

Conversation based intelligent search enabling customers to look for apparels in retails store.dialogue management using Rasa core/NLU with w2v embeddings enabled search api

automobile-price-prediction

figure out most influencing underlying factors and design a regression model to predict car price for a chinese automobile company

chest-xray-classification

identify effusion in chest Xray images using morphological transformation with Opencv for image processing and 2D-CNN for image classification.

demand-forecasting-framework

detecting-vehicles-in-video

this project does counting and classification of vehicles passing on a road using opencv based blob feature extraction to do vehicle detection and CNN based vehicle classification

exam-proctoring-video-analytics

This project implements a computer vision based virtual online exam proctoring software by capturing facial recognition, head pose and eye gaze through webcam using CNN based deep learning models

facebook-friend-recommendation-using-graph-mining

used Random Forest classifier to predict missing links to recommend users in a social graph generated from historical data using probabilistic grahical model

gesture-recognition

developed a 3d-cnn and conv2d(pretrained-inception)-gru to correctly recognize hand gestures by a user to control a smart TV.

hh-reco

human-activity-recognition-from-mobile-data

Human activity recognition is the problem of classifying sequences of accelerometer data recorded by specialized harnesses or smart phones into known well-defined movements.LSTM gave best results

image-synthesis

lending-club-case-study

Carry out EDA using data visulaization techniques to figure out driving factors (or driver variables) behind loan default for a consumer finance company

llm

llm-rag-prompts

microsoft-malware-detection

to determine malwares in bytes/asm file ,carried out multivariate EDA using t-sne and implemented XgBoost Classifier on final features with best hyper parameters using Random search for classification

movie-recommendation

movie recommendation with RL

netflix-movie-recommendation

created matrix factorization (SVD) based collaborative filtering algorithm to recommend movies to netflix users based on historical preferences

nlp-1

nyc-taxi-prediction

Taxi demand prediction on hourly basis for Ny city.The modelling is done on moving average and fourier transform based features after applying tsne .xgboost gave best performance

paintshop-process-optimization

paitshop optimization

personalized-cancer-detection

Classify the given genetic variations/mutations based on evidence from text-based clinical literature.TF-IDF features along with Random Forest Classifier gave best results

quora-question-pair-similarity

To filter similar question asked in popular QA site,tried out models (Logistic regression, Linear-SVM) with TD_IDF weighted word2Vec. Hyperparameter tune XgBoost using RandomSearch to reduce the log-loss.

reinforcement-learning--cab-driver-optimum-ride-selection

developed RL-based algorithm using state based DQN architecture with simulated MDP to help cab drivers effectively decide which cab-request to accept for maximum profit

restaurant-search-chatbot

developed conversational bot (chatbot)to help users discover restaurants quickly using Rasa Core and NLU framework with zomato api

retail-demand-forecasting

Demand Forecasting for weekly storewise and productwise sales has been projected for a global retailer.An ensemble of xgboost and catboost regressor is done on lag features to arrive at final output

rl-demand-forecasting

demand forecasting with RL

rl-recommendation-engine

rl-recommender

rl-timeseries-forecasting