Portfolio

Cloud Token

A side project of mine which involved developing an ERC-20 token, which is a fungible asset on the Ethereum blockchain. Seemingly unrelated to Data Science but, nonetheless a project I undertook out of curiosity and interest. Building immutable applications on the Ethereum blockchain has many usecases, but this was my attempt at building something solo. The token is available on the Rinkeby Test Net at address
0x02c3e2E97b3a3E05dDc78a165af72674b51B8155.

Marketing Analysis

This project involved customer segmentation and forecasting the profitability of customers. Interestingly this was a multivariate time series forecasting project, with a mixture of unsupervised learning due to the use of clustering algorithms. The end result was a model which was delivered to the marketing division of my client, along with a presentation to key stakeholders pointing out important information discovered during the project.

Realty Forecast

One of my favorite projects, this one involved time series forecasting the median market value of houses by zipcode in the USA. I was tasked with identifying market opportunities for a real estate company seeking to expand its portfolio to other regions in the USA. Using SARIMA/ARIMA modeling, I was able to identify real estate opportunites for my client in the state of Colorado which in a year I projected would be profitable (with 95% confidence).

Customer Churn

This project tested my ability using classification algorithms, and required me to identify whether a customer would soon terminate their service plan with a telecommunications company. Using data provided by the company, I was able to generate a model with an accuracy of over 95% (based on holdout validation). Through modeling I was able to identify the main reasons why customers terminated their plans, and presented this vital information to key stakeholders.

Housing Market

This project was devoted entirely to regression (specifically multivariate linear regression). I was tasked with generating a regression model which would take as input the features of a house, and output the approximate market value. This model would then be used to identify which features have the greatest impact on the market value of a home in the region.

Movie Industry

This was my first data science related project, and entirely focused on the Exploratory Data Analysis phase of the Data Science Life Cycle. I was tasked with identifying what type of movies Microsoft should produce, for their ficticious new movie studio. It culminated in a non-technical presentation to key business stakeholders.