Himalay Patel's repositories
credit-card-fraud-detection
Implemented Decision Tree, Random Forest and Deep Neural Network (5 layers) on a real anonymized dataset. Used undersampling and oversampling (SMOTE) with Deep Neural Network to balance the highly imbalanced dataset (very less fraudulent transactions as compared to non-fraudulent ones). Achieved 99.6% test accuracy using SMOTE with NN along with 0 Type-II errors (false negatives).
minimizing-churn-rate
Applied Logistic Regression to a dataset of customers’ financial habits artificially created from real life case studies. Achieved test accuracy of 62.9%. Used Undersampling to balance the dataset, k-fold cross validation to improve accuracy and Recursive Feature Elimination to reduce chances of overfitting.
angular.js
AngularJS - HTML enhanced for web apps!
AutoRCCar
OpenCV Python Neural Network Autonomous RC Car
customer-subscription-direction
Used Logistic Regression Classifier to decide whether a customer should be directed to subscription by analyzing his app behavior. Also implemented k-fold Cross Validation to prevent overfitting.
home-security-system
Built system to recognize faces and give access to home, based on access profile (family, friend, intruder) in database. Possesses multi-level security. (PIR motion sensor inside for motion detection and webcam outside home for surveillance and face recognition) Owner/s can access system remotely and can also see live video surveillance on mobile phones/PC at any time. (Check out test_videos.txt for links to YouTube videos to see the project in action!)
loan-likelihood-prediction
Implemented following algorithms to solve the problem- Logistic Regression (with Lasso penalty), SVM (with linear and RBF kernels) and Random Forest (with 100 decision trees and Entropy & Gini impurity). Also used Grid Search to tune hyperparameters of Random Forest classifier. Achieved initial test accuracy of 56.2% with Logistic Regression and increased to 63.5% with Random Forest + Gini impurity.
mnist-digits-classification
Built a linear neural network from scratch to classify handwritten digits (0-9) taken from MNIST datasubset in json format. Implemented Mini Batch Gradient Descent, ReLU activation function, softmax for classification and dropout to handle overfitting. Achieved test accuracy of 92.6%.
awesome-courses
:books: List of awesome university courses for learning Computer Science!
awesome-public-datasets
A topic-centric list of HQ open datasets. PR ☛☛☛
Babylon.js
Babylon.js: a complete JavaScript framework for building 3D games with HTML 5 and WebGL
blynk-library
Blynk library for embedded hardware. Works with Arduino, ESP8266, Raspberry Pi, Intel Edison/Galileo, LinkIt ONE, Particle Core/Photon, Energia, ARM mbed, etc.
breast-cancer-classification
Used Support Vector Machine to classify a given image from breast cancer wisconsin dataset (sklearn) into malignant or benign cancer.
ember.js
Ember.js - A JavaScript framework for creating ambitious web applications
face_detect_n_track
Fast and robust face detection and tracking
face_recognition
The world's simplest facial recognition api for Python and the command line
fashion-class-classification
Implemented Convolutional Neural Network to classify images from MNIST Fashion Class dataset into 10 different classes.
linux
Linux kernel source tree
meteor
Meteor, the JavaScript App Platform
milq
Useful code of Manuel Ignacio López Quintero
opencv
Open Source Computer Vision Library
rails
Ruby on Rails
react
A declarative, efficient, and flexible JavaScript library for building user interfaces.
remi
Python REMote Interface library. Platform independent. In about 100 Kbytes, perfect for your diet.
swot
Identify email addresses or domains names that belong to colleges or universities. Help automate the process of approving or rejecting academic discounts.
toy-autonomous-cars
Implemented value iteration algorithm and generated an optimal policy for cars moving in an observable, stochastic and reward-based environment. Also simulated the policy generated introducing stochasticity using numpy random and calculated the average rewards for each car.
vue
A progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Vulcan
🌋 A toolkit to quickly build apps with React, GraphQL & Meteor