dplyr, package creation, data wrangling, flight map using Nathan Yau's code

Neural Nets using R. Answer to STATS 315B homework question.

Ensemble learning using logistic, boosting, and SVM.

Principal Components Regression and Partial Least Square Regression

Ridge and Lasso Regressions.

In which I found out that non-linearity in Sib/Spouse variable is HUGE! It's not overfitting either because I found that adding this factor to the training set helps and then it significantly improved on predictive power on the test set. But WHY?

Include initial data cleaning, adding some new variables. Used Best Subset Selection method to pick Variables but didn't do full CV.

SVM using linear and radial kernels.

Using Random Forest and Boosting. Boosting performed substantially better!

This is first attempt at creating models, making prediction and submitting results to Kaggle's March Madness competition. Following a tutorial from a blog.

Clustering when there are none.

Anyone who considers arithmetical methods of producing random digits is, of course, in a state of sin (John von Newman)

My answers for HW4.

From FlowingData tutorial. Will try this for Hong Kong.

From "ML for Hackers" book. Chapter 3. Most critical step is learning how to build a "Term Document Matrix": Columns contain all the terms found in all of the documents. Rows are the emails (each document). tm package is very handy for text distilling.

From Lab 10 in ISLR. Using Unsupervised Learning techniques - PCA and Hierarchical Clustering.

Testing out R Presentation with this analysis

R-lab from SVM lecture in Stats 216

In-class example of Linear Regression Analysis for NCAA basketball data

Homework 1 for Stats 216 Stanford

Homework 2 for Stats 216 Stanford

My homework 3 answers for Stats 216 Stanford