Fast Stats About Me
- BA: Economics - University of Iowa
- MS: Management of Information Systems - University of Illinois
- Professional Certificates:
- Artificial Intelligence and Deep Learning - DePaul University
- Focus Areas: Marketing Analytics, Predictive Modeling & Machine Learning
I am a data analyst based in Chicago that specializes in digital media and web analytics. I have a background in eCommerce analytics spanning industries across different B2C and B2B verticals.
Areas of expertise include:
- Demand Forecasting
- Churn prediction (e.g. are we emailing too frequently?)
- Lift analysis (e.g. does Display work? Is it worth bidding on my own brand?)
- Paid Search Analysis
- A/B Test Design
- Google Tag Manager implementations
- Google Analytics advanced segmentation and reporting
- Tableau dashboarding
These are just a subset of common areas I frequently find companies need help with. Through the use of proven statistical analysis and an understanding of business problems it's often non-trivial to get to the root of the problem and extract insights relevant to improving revenue or business process.
Attribution using Probabilistic Graphical Models
This is a project I started to leverage probabilistic graphical models to solve for attribution modeling problems in marketing and advertising. While versions of this exist in R, I have yet to see one in Python. It currently provides a solution for modeling existing, sequential, user paths' probabilities and calculating the contribution of each channel by examining the affect or removing a node (a channel in this instance) entirely from the graph. You can read more here or jump straight to the code here.
Dynamic Classification of Images of Clothing
This project was started after exploring deep learning and the use of convolutional neural networks to recognize images. I wanted to practice deploying a neural network to a cloud provider (GCP in this instance) to mimic a production scenario that would allow for users to interact with the model to classify their own sets of images based on the model I had trained. Current status is a command line app that can recognize images fed to it. I still need to add the resampling code to allow for any shape or size of images to be tested against the model.