Matthew Merrill

Repository for testing my Git/GitHub setup

My github profile page

Created prediction algorithm for determining if a customer will cancel at the moment of booking. After eliminating numerous data leakage sources, we achieved a 90% AUROC with a catboost classification model.

Recommendation engine with a .97 AUC achieved using clustering techniques to create user features. Data represents Olist marketplace transactions and was retrieved from kaggle.com.

A slide deck and set of jupyter notebooks created during my time with Fellowship.ai in order to decide what method would be best for predicting churn using the most up to date and innovative methods.

📝 Easily create a beautiful academic website using Hugo, GitHub, and Netlify

Identified the most important factors contributing to user adoption for a product. Achieved a 96% accuracy with an SVM model, and found user length and opting into the mailing list as the most significant predictors.

No description provided.

A repository of projects completed through the Springboard career track program.

My first R Shiny app for exploring the catboost model for predicting hotel cancellations.

My portfolio website. Created using the academic template through wowchamy.

Project aimed at differentiating between positive and negative reviews using the fastai's ULMFiT implementation method.

Recommendation system dockerized using a lightfm image.

Built and deployed docker container using Kubernetes KNative manifest for full on serverless machine learning deployment.

https://share.streamlit.io/merrillm1/predicting_cancellations_streamlit_app/predicting_hotel_cancellations.py

With the continuing drop in rent prices accross San Francisco, and with the financial strain that transitioning careers has left me with , I decided to put my Data Science skills to good use and ask for an evidence based proposal of lowering my rent.

Predicted rider retention for a taxi service and identified most significant factors that contributed to it. Achieved an 80% accuracy with a catboost model, which was chosen for its interpretability.