Experiments were carried out with a group of 30 volunteers within an age bracket of 19-48 years while wearing a smartphone (Samsung Galaxy) on the waist Each person performed six activities:
- Walking,
- Walking upstairs
- Walking downstairs
- Sitting
- Standing
- Laying
Reference: https://www.kaggle.com/datasets/uciml/human-activity-recognition-with-smartphones
The data collected consists of 561 different features generated from the raw accelerometer and gyroscope signals.
Developed two predictive models, Neural Network and SVM using all 561 features.
For each model, reported:
1- Accuracy
2- Confusion matrix
For SVM model: Varied the number of features using PCA and computed the resulting accuracy
Determined the number of features required to obtain 80% , 90% accuracy