Human Activity Recognizer

This repository contains a project to recognize human activities using machine learning techniques. The model is trained and tested on the UCI Human Activity Recognition dataset. The pipeline involves feature selection, classifier training, and testing.

Files and Directories

• HAR.ipynb: Notebook containing the training code, including data preprocessing, feature selection, and model training.
• test.ipynb: Notebook used to test the trained model on new or unseen data.
• utils.py: Python script that contains utility functions such as removeDuplicateColumns, which is used to clean the dataset by removing redundant features.
• model.pkl: Trained Random Forest model saved in pickle format using joblib, which can be loaded and used in the test.ipynb for making predictions.

Dataset

The dataset used in this project is the UCI Human Activity Recognition dataset, which can be downloaded from this link.

Approach

The model training approach is organized as follows:

1. Label Encoding

• The activity labels (output) are encoded into numerical values to be used by the classifier.

2. Removing Duplicate Columns

• The dataset is checked for redundant columns, which are removed using the removeDuplicateColumns function from the utils.py file.

3. Feature Selection using SelectKBest

• SelectKBest is applied to select the top k most important features based on statistical tests.

4. Recursive Feature Elimination (RFE) with Random Forest Classifier

• Recursive Feature Elimination (RFE) is applied using the Random Forest Classifier to further refine the feature selection and remove the least important features iteratively.

5. Random Forest Classifier

• A Random Forest Classifier is trained using the selected features. Random Forest is chosen due to its robustness and ability to handle complex datasets effectively.

Pipeline

• Steps 2 to 5 are organized into a pipeline, ensuring a streamlined and reproducible process for data preparation, feature selection, and model training.

Exporting the Model

• The trained model is exported into a .pkl file using joblib for later use in the test.ipynb notebook for testing and predictions.

How to Run

Training the Model

1. Open and run HAR.ipynb to:
   • Preprocess the data.
   • Train the Random Forest model.
   • Export the trained model to a .pkl file.

Testing the Model

1. Open and run test.ipynb to:
   • Load the saved model (pipeline.pkl).
   • Test the model on new data or the test set.

Requirements

Make sure to install the following dependencies:

pip install numpy pandas seaborn matplotlib scikit-learn joblib