How to Run the Project ▶️

To execute the project, follow these steps:

1. Create and activate a virtual environment:

   • Navigate to the project folder (if not already there):

     cd final_project/

   • Create a virtual environment:

     python -m venv myenv

   • Activate the virtual environment:
     • On macOS/Linux:

       source myenv/bin/activate

     • On Windows:

       myenv\Scripts\activate

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   streamlit run app.py

Project Context 📚

This project is part of an investigation into “Intelligent Decision-Making with Machine Learning using Explainability Techniques for Time Series Problems.” The work was supervised by Professor Fábio Silva and aimed at researching machine learning models for anomaly detection in time series data.

Objectives 🎯

The primary goal of this project is to enhance the explainability of machine learning models applied to anomaly detection in time series. Specific objectives include:

• Studying time series data.
• Exploring Explainable AI (XAI) frameworks.
• Analyzing black-box and white-box models.
• Developing proof-of-concept implementations.
• Conducting a case study using 5G network datasets.

Results 📊

The project involved training models such as Naïve Bayes and Decision Tree on pre-processed datasets. Explainability techniques like SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-Agnostic Explanations) were applied to interpret the decision-making process of these models.

Additionally, a web platform was developed using Streamlit, allowing users to:

• Import datasets and visualize time-series data through interactive dashboards.
• Analyze trained models and their explainability outputs.
• Generate predictions with trained models and view LIME-based explanations for each prediction.