ActivitySense: Classification of Human Activities Using Smartphone Sensor Data

This repository contains code and analysis for a simple machine learning project to classify human activities using smartphone sensor data.

Overview

ActivitySense is a project focused on developing a machine learning model for accurately classifying human activities using smartphone sensor data. The project aims to categorize activities such as walking, sitting, standing, and more based on recordings from individuals performing various activities of daily living.

Dataset

The dataset is from [1] (Reyes-Ortiz,Jorge, Anguita,Davide, Ghio,Alessandro, Oneto,Luca, and Parra,Xavier. (2012). Human Activity Recognition Using Smartphones. UCI Machine Learning Repository. https://doi.org/10.24432/C54S4K.). It contains recordings of 30 participants performing activities of daily living while carrying smartphones.

Project Components

1. Motivation: Discusses the importance of accurately classifying human activities from smartphone sensor data in industries like fitness tracking and healthcare.
2. Objectives: Outlines the primary goals of the project, including developing a machine-learning model to classify activities into specific categories.
3. Methodology: Describes the approach taken in the project, including class distribution in the dataset, feature engineering techniques, and the implementation of various machine learning algorithms.
4. Experiments and Results: Presents the comparison of execution time, evaluation metrics for original and PCA reduced data, and the conclusions drawn from the results.
5. References: Lists the sources and research papers referenced in the project.

How to Use

1. Dataset: The project utilizes a dataset of recordings from individuals performing activities of daily living captured through wearable smartphones. Ensure you have access to this dataset for replication.
2. Code: The project involves data preprocessing, exploratory data analysis, feature engineering, and the implementation of supervised learning algorithms. The code files are structured accordingly.
3. Results: Review the experiment results to understand the performance of different machine learning algorithms in classifying human activities.
4. Future Work: Consider exploring additional feature engineering techniques and fine-tuning hyperparameters for further improving the classification model.

Technologies

Python, NumPy, Pandas, Scikit-Learn, Matplotlib

Acknowledgments

This project is part of the Master of Computer Engineering for Robotics and Smart Industry program.