Project information

• Extracted practicable features from raw time series gyroscope readings using data segmentation and windowing, and used Principal Component Analysis (PCA), Feature Importance etc. to select the minimal subset of features for accurate performance of the ML models.
• Built an API to be used by mobile phones to post streams of real time gyroscopic readings, and get the classifier results.
• Built a mobile application to bridge the interaction between users and the web API, and to stream real time gyroscope readings from mobile devices.

ML, DL, Python, TensorFlow, JavaScript, No-SQL, MQTT