MLB Pitch Outcome Classification Capstone

Overview

This project is the final/Capstone project for Flatiron School's bootcamp program in Data Science. We have created a hypothetical situation as a Data Scientist and are hoping to provide value to our business for the scenario.

Business Problem

We have been hired as a hypothetical member of the Cardinals baseball organization: a member of the coaching staff. As a coaching analyst, our job is to create a model that will give us insights into pitch quality and classify a pitch, given its metrics, as a negative, neutral, or positive outcome for the pitcher.

Project Deliverables

A GitHub repository
A Jupyter Notebook
A non-technical presentation

Project Summary

I devised these questions that I believed could be answered through data analysis.

1. What are the most important metrics that go into a pitch?
1. What is the least important metric that goes into a pitch?

I explore this thoroughly in the pitch-classification.ipynb file contained within this repository via classification modeling. I progressively alter the scope of the models as I iterate over different options. Here are some visualization previews of the data I investigated.

Corr Heatmap Between Features and Pitch Outcome

Single Player XGB Model Feature Importances

Entire Team XGB Model Feature Importances

Repository Structure

├── images
├── README.md
├── pitch-repo.pdf
├── pitch-presentation.pdf
├── pitch-notebook.pdf
└── pitch-classification.ipynb

timothyf / pitch-analysis