Decision Tree Model for Room Locationing based on Wi-Fi Reception

This project provides a decision tree-based model for predicting room numbers based on the receiving Wi-Fi signal strengths.

Prerequisites

Before running the code, ensure you have the following dependencies installed on the Lab Machines:

• Python 3.10.12
• NumPy 1.24.3
• Matplotlib 3.7.2

To install the dependencies, run:

pip install matplotlib==3.7.2 numpy==1.24.3

Data

Make sure you update the __main__ section of the decision_tree.py file:

# Main for script execution:
if __name__ == '__main__':
    dataset_path='wifi_db/clean_dataset.txt'
    model = DecisionTreeModel(dataset_path, folds=10)
    model.run()

• dataset_path : File path of the dataset to load and run the program on.
• folds : Number of cross validation folds to perform on the dataset.

How to Run

1. Download the file decision_tree.py to your local machine.

2. Update the __main__ as required with the dataset's file path.

3. Execute the main script:

   python decision_tree.py
   

This will do the followings:

• Build a Decision Tree Model on the entire dataset.
• Run cross validation based on the choosen number of folds.
• Generate necessary plots and compute various cross validation classification metrics.

Outputs

Upon successful execution, you will get:

1. A visualization of the Decision Tree Model trained on the full dataset as Tree.png.
2. A confusion matrix saved as Confusion_Matrix.png generated from the k-fold cross validation.
3. A plot representing loss vs. depth of the tree saved as Loss_vs_Depth.png showing data-subsets become purer with increasing tree depth.
4. The maximum depth of the initial Decision Tree Model produced, printed on console.
5. The accuracy of the algorithm computed from k-fold cross validation, printed on console.
6. Cross validation classification metrics such as recall, precision and F1-measure for each class, printed on console.
7. The macro-averaged values of the above metrics from each class, printed on console.

The .png images produced will be saved in the same directory where the script is being executed.

---