Data Mining and Machine Learning @ CEID

Technologies

NumPy, Matplotlib, Pandas, scikit-learn, Pytorch, pgmpy, SciPy, Seaborn

Dataset

Human Activity Recognition Trondheim (HARTH) The HARTH dataset, contains data collected from two accelerometers worn by 22 participants in a study. The data collection lasted approximately 2 hours in a free-movement environment.
Performed an initial assessment of the data, followed by classification based on activity using

• Neural Network implemented in PyTorch,
• Random Forest from scikit-learn,
• Bayesian Networks in pgmpy & Bayesian Classifier from scikit-learn.
  
  
  Afterward, K-Means clustering was applied to group the data according to the detected activities.