Machine-Learning
This repo contains my projects on Machine Learning, I have covered diverse topics under each project, You can see the description for each project below.
1)Classify Song Genres from Audio Data
This project is based on recommending new music to the users by analysing various components. I have used PCA in this project for dimensionality reduction. For modelling I have tried different models like Decision Tree and Logistic Regression. I have further used Corss validation to evaluate the models performance.
2)Give Life_ Predict Blood Donations
This project is based on binnary classification where I predict if a donor who has donated blood in previous 6 months will dontate blood again or not. It was fun as I got to implement pipelines in this project which was something new for me.
3)Prediciting Titanic Survival
In this project I worked on a Kaggle data set (Titanic Data Set). Based on all the features available I tried predicting whether a passanger will survive or not. I was able to land in top 14% of the competetion. I have used XGBoost along with hyper parameter tuning to get my results.
Many American cities have communal bike sharing stations where you can rent bicycles by the hour or day. Washington, D.C. is one of these cities. In this project, I tried to predict the total number of bikes people rented in a given hour by using variables like monthly rental, weekly rental, temperature, humidity etc. We used Linear Regression as well as Decesion trees and compared which model provided us better results.
In this project I tried predicting price of car based on training my model with attributes like fuel-type,engine-type,compression-rate,horsepower etc. I have used KNeighbors as my model to predict the car price.
6)Predicting Credit Card Approvals
I worked on this project to get a feel of fraud detection analysis. From the given dataset I tried to predict whether a loan should be approved or not based on various features which were available. I used logistic regression as my model and I also used hyper parameter tuning to enhance the performance of my model.
7)Predicting board game reviews
In this project I worked with a data set that contains 80000 board games and their associated review scores. We used various parameters of review and tried to predict the average rating of a board game. We used co-relation to find out relavant variables for my analysis and Linear regression for modelling.
8)Reducing Traffic Mortality in the USA
This project was based on the increasing rate of road accidents. I used unsupervised learning for training my model and PCA for dimensionality reduction. I also tried to display a concept called as masking by using multivariate regression.
9)Predicting TMDB Box Office Collections
In this project, I tried to predict box office collections of movies based on various features provided in the data set. It was a very intersting project to work on as the data format was something new to me and I got to learn and explore new dimensions of Data analytics.
This is a kaggle competition I worked on which had only 250 rows in train set and around 1000+ rows in test. The task was to predict the test set without overfitting the training set. I have used various models like xgboost, linear regression and later combined them inorder to achieve better performance.