IML Work 4

Forth work of UB course "Introduction to Machine Learning" implementing dimensionality reductions and visualization techniques

Names of contributers

Eva Veli, Andras Kasa and Niklas Long Schiefelbein

Table of Contents

• Project Setup Instructions
  • Manual Virtual Environment Setup
  • Installing Dependencies
  • Close Virtual Environment
• Project Execution instructions
  • Make sure to be in the root directory IML4
  • Activate the Virtual Environment
  • Run app.py
• Project Structure

Prerequisites

• PyCharm IDE (Professional or Community Edition)
• Python 3.9 installed on your system

Project Setup Instructions

1. Manual Virtual Environment Setup

1. Open the project IML4 in PyCharm

2. Open the terminal in PyCharm (View > Tool Windows > Terminal)

3. Optional: Verify current location being IML4 by pwd

4. Optional: Navigate to IML4 with cd

5. Create a virtual environment:

   # Windows
   py -3.9 -m venv venv
   
   # macOS/Linux
   python3.9 -m venv venv

6. Activate the virtual environment:

   # Windows
   venv\Scripts\activate
   
   # macOS/Linux
   source venv/bin/activate

In front of the input line in the terminal it should now say (venv)

2. Installing Dependencies

With the virtual environment activated:

pip install -r requirements.txt

From here you can directly jump to Run app.py

3. Close Virtual Environment

With the virtual environment activated:

deactivate

The (venv) in front of the terminal should be gone

Project Execution instructions

1. Make sure to be in the root directory IML4

For this, just follow the optional steps 3 and 4 from the Manual Virtual Environment Setup

2. Activate the Virtual Environment

# Windows
venv\Scripts\activate

# macOS/Linux
source venv/bin/activate

In front of the input line in the terminal it should now say (venv)

3. Run app.py

python app.py

The first execution takes more time than usual due to the initial compilation of the whole project. Once compiled, it prompts the user to provide an input. The user must decide whether to use the cmc or the pen-based dataset for the analysis. By simply pressing enter, the cmc dataset will be selected by default.

Now the preprocessing pipeline will execute. After that the user can choose from 11 menu items including executing the clustering algorithm, report generation and exiting the program. Menu point 1 includes all PCA-Analysis items. Progress for each functionality is displayed in the console, but due to the fast computation it may be difficult to follow at all times. It is recommended to refer to the final reports for evaluation.

For deeper insights please consider reading the report of the project.

Project Structure

IML4/
├── cluster_algorithms/    # K-Means++ and OPTICS
├── datasets/              # Dataset files
├── metrics/               # Distance and evaluation metrics
├── pca/                   # PCA code and projection data
├── plots/                 # All plots
├── plotting/              # Code used to create plots
├── preprocessing/         # Preprocessing functions
├── results/               # CSV results for each algorithm
├── summary/               # LaTeX tables and their code
├── .gitignore             # Gitignore file
├── app.py                 # Main application script
├── README.md              # This file
├── requirements.txt       # Dependencies
└── utils.py               # Utility functions