CS229 Predicting Heart Attacks

Plead or Pitch? Predicting the Performance of Kickstarter Projects

In this CS229 project, using 26K project proposals from crowdfunding platform Kickstarter, we evaluate the performance of different models (logistic regression, SVMs among others) on predicting whether a project will meet its funding goal or not, with a particular focus on features derived from the language used in project pitches. We then contrast the performance of our best model when using d...

Using Pre-NBA Draft Data to Project Success in the NBA

The NBA draft poses a unique challenge in that predicting a player’s future success in the NBA is incredibly difficult. We seek to use machine learning techniques to quantify the attributes that tend to indicate a college player’s playing ability in the NBA. Using historical data from players’ college careers in combination with their NBA career data, we have developed a model to predict where ...

Predicting coronary heart disease from minor symptoms and electrocardiographic findings.

Many coronary heart disease (CHD) deaths and many attacks of myocardial infarction seem clinically to occur as unheralded catastrophes. This is surprising, considering that pathologically such events are preceded by years of progressive atherosclerosis and ischaemia. The present study estimates the frequency with which minor symptoms and electrocardiographic changes may occur in the years befor...

Predicting Popularity of Posts on Hacker News CS229 Autumn 2016 Project Final Report

In this project, we try to find the best popularity predictor on the dataset of Hacker News posts. Features including n-gram, simple counts, tf-idf, word2vec and topic models and classifiers like SVM, Naive Bayes, etc are experimented. Data resampling is applied to combat the imbalance of the dataset. By combining different features, classifiers and resampling techniques, we are able to find pr...

CS229 Project: Segmenting Music into Notes using Mixture of Gaussians