icodein / codeportfolio

Code Portfolio

The following are assignments and projects completed during my time at Northwestern University. Below are brief descriptions of each folder's content.

Big_Data

Contains code for Hadoop, Hbase, and Hive assignments with the PDFs describing the assignment details. Some of the underlying data is not included given the size of the datasets, which was stored in our school's cluster.

Data_Visualization

Created a social network visualization of characters in a popular online game called Pokemon Showdown. Includes Python code for preprocessing data from multiple data sources, jsons for node/edge data, CSS code for html formatting, R code to integrate network statistics, and d3.js code to create the visualization itself of a force directed graph. Also includes a live link to the final visualization in the readme.

Deep_Learning

All code was created with Python Keras/Tensorflow and run on our university's GPU cluster. Assign 1 and 2 are labs we conducted in class to learn about neural network tuning and convolutional neural networks. The project folder contains code for a convolutional neural network we created to train an A.I. to play an online arcade game called Rumbah. The ppt contains a good description of our overall project.

ERGM

This was a lab I completed for my social networks class where we used igraph to analyze a communication network in a team combat simulator. We utilized an Exponential Random Graph Model to analyze network attributes that are most significant and the docx provides a good description of the overall assignment and visuals.

Java_Data_Structures

This was from a class I actually took prior to Northwestern University where I learned about Data Structures using Java.

Pred_Analytics_Project

From one of my first classes at Northwestern, my team was given a dataset from a nonprofit organization. We utilized linear and logistic regression to find the probability that a person donates and in what amount. Results were matched against a test data set for top 1000 donors.

Statsllc_Mclust

Worked with Stats LLC to use unsupervised techniques such as Explanatory Factor Analysis and Gaussian Mixture clustering to find groups of NBA players by skillset.

Python_Java

A collection of Python and Java assignments we had for a programming course. PDFs describe the assignment parameters.

Pred_Analytics_2

A collection of assignments we completed studying advanced tehcniques such as decision trees, random forests, boosted trees, cross validation, bootstrapping, GAM, and time series models.