Project information

• Category: Machine Learning
• Professor: Nipun Batra, IITGN
• Project date: Jan '24 - Feb '24
• Project URL: ML_MiniProject_HCI_Dataset

This project focuses on developing a Human Activity Recognition (HAR) model using machine learning techniques to classify different human activities based on sensor data from the HCI (Human-Computer Interaction) dataset. The process involves data preprocessing, feature extraction using the TSFEL (Time Series Feature Extraction Library), and dimensionality reduction with Principal Component Analysis (PCA). Various machine learning models, including Decision Trees, are trained and evaluated using metrics such as accuracy, precision, recall, and confusion matrices. The project highlights the importance of data preprocessing and feature extraction in improving model performance, with TSFEL significantly enhancing the classification accuracy. The final model is capable of classifying activities in real-time, demonstrating the practical application of machine learning to real-world sensor data. The project concludes that dynamic activities are classified more accurately due to higher variability in the data, while static activities pose challenges due to less variability.