---
title: "Projects"
description: "Side projects by Darryn Pasqua."
canonical: "https://darrynpasqua.com/projects/"
---

# Projects

Sports analytics · machine learning 

 Pitch Classifier 
 Completed 

 Neural network workflow for classifying MLB pitch types from 2024 Statcast data, with separate models for left- and right-handed pitchers. 

 Left-handed pitchers 
 94.67% 
 best validation accuracy 

 Right-handed pitchers 
 92.78% 
 best validation accuracy 

 30-day Statcast pulls 
 18 engineered inputs 
 Separate LHP / RHP models 

 Context 
 INSY 7130 · Auburn · 2024 

 Best validation 
 94.67% LHP · 92.78% RHP 

 Feature set 
 18 release, movement, trajectory inputs 

 A baseball analytics project that downloads Statcast pitch data, tunes a feed-forward neural network, and trains separate classifiers for left- and right-handed pitchers using release, movement, and trajectory features. 

 Python 
 TensorFlow 
 Keras 
 KerasTuner 
 scikit-learn 
 pandas 
 pybaseball 
 matplotlib 

 More detail 

 Downloads 2024 MLB Statcast pitch-level data in 30-day chunks. 
 Tunes the model with Hyperband before final training. 
 Trains separate classifiers for left-handed and right-handed pitchers. 
 Includes plots, saved histories, model artifacts, and the full report. 

 Repository 
 Project report 

 Fitness software · desktop app 

 Bike Tracker 
 Ongoing 

 Real-time indoor cycling analytics desktop app. 

 Role 
 Python desktop app for live FTMS bike telemetry 

 Core loop 
 Bluetooth metrics in, ride analytics and session history out 

 Storage 
 SQLite-backed workout records for later review 

 A Python app that connects to a Bluetooth FTMS exercise bike and turns live ride data into a usable training dashboard. It shows cadence, power, heart rate, speed, resistance, VO2 estimate, and training stress in real time, while recording sessions to SQLite for later review. 

 Python 
 PyQt6 
 BLE 
 SQLite 
 pyqtgraph 

 View Project 
 See on GitHub