User Activity Recognition

About

This project is an User Activity Recognition basing on Mobile Devices, we do some analysis experiment using Python, and develop the pipeline using Java to support Android system.

Description

• Folder Preprocess corresponds to transforming JSON format data to a tab-separated String
• Folder ExtractFeature corresponds to extracting features and saving results to an arff file
• Folder Classifier corresponds to implementing classifier, which include traning, testing and serializing model into file
• Folder ResultValidate corresponds to smoothing the prediction result to generating trace of activity for each user
• The other folders corresponds to some auxiliary tools

Tools

• feature_selection_UAR.ipynb contains the process of feature extraction
• curve_visualize.ipynb contains visualization form of dataset

DataSet

• dataset contains the whole training dataset
• data_for_analysis contains partial dataset for analysis