Machine Learning Activity Classifier

Academic Project · Desktop Application

Feb 2025 – Apr 2025

Developed a Tkinter desktop application built with Python. The interface allows uploading accelerometer data collected from a phone through a CSV file. Once a file is loaded, the graph updates to show the raw X, Y, and Z acceleration over time.

To classify the motion, the model breaks the data into five-second windows, extracts statistical features, scales them, and applies a trained logistic regression model to label each segment. The predictions are exported as a CSV, showing whether the activity in each window was walking or jumping.

Skills: Python · Tkinter · Machine Learning · Data Science