MLB_EWHA Machine Learning Project

This repository contains the implementation of the Machine Learning course project. The project explores Closed-world and Open-world classification tasks using datasets of monitored and unmonitored websites.

---

📂 Repository Structure

• mon.csv: Features extracted from the mon_standard.pkl file.

• unmonmon.csv: Features extracted from the unmon_standard10_3000.pkl & mon_standard.pkl file.

• load_pickle_code.ipynb: The datasets were preprocessed and converted into features.

• Task Process/: Includes all the code used during project development.

• Optimized Models/: Contains optimized code for evaluation purposes.

---

📊 Project Overview

1️⃣ Closed-World Classification

• Objective: Classify 95 monitored websites into 95 unique labels.
• Dataset: mon.csv (processed from mon_standard.pkl).

2️⃣ Open-World Classification

• Objective: Include both monitored and unmonitored datasets for the following tasks:
  • Binary Classification: Predict whether a trace is monitored (1) or unmonitored (-1).
  • Multi-Class Classification: Classify 95 monitored websites (0-94) and unmonitored websites (-1) into a total of 96 classes.

---

📚 Datasets

Download Links:

• mon_standard.pkl
• unmon_standard10_3000.pkl

Dataset Details:

Dataset	Instances	Classes	Description
mon_standard.pkl	19,000	95 (0-94)	Monitored website traces. Each website has 10 subpages, observed 20 times each.
unmon_standard10_3000.pkl	3,000	(unmonitored only)	Unmonitored website traces.

---

🚀 Setup and Execution

1️⃣ Environment

The code was executed in the following environment:

• Python Version: 3.8.10

2️⃣ Required Libraries

To run the code, the following Python libraries need to be installed:

• pandas
• numpy
• matplotlib
• scikit-learn
• xgboost
• imbalanced-learn

3️⃣ Steps to Run

1. Clone the repository and navigate to the project directory.

   git clone https://github.com/MLB-EWHA/MLB_TEAM.git
   cd MLB_TEAM

2. Install the required libraries listed above.

   pip install pandas numpy matplotlib scikit-learn xgboost imbalanced-learn

3. Run the desired classification tasks:

   • Navigate to the Optimized Models/ folder.

   1. Closed-World Classification:

      • Run the script:

        Closed_Multi_Random Forest.ipynb

      • This script builds a model to classify the 95 monitored websites into unique classes.

   2. Open-World Binary Classification:

      • Run the script:

        Open_Binary_XGBoost.ipynb

      • This script builds a binary classification model to distinguish monitored (1) from unmonitored (-1) website traffic.

   3. Open-World Multi-Class Classification:

      • Run the script:

        Open_Multi_Random Forest.ipynb

      • This script builds a multi-class classification model to classify 95 monitored websites (0-94) and unmonitored websites (-1) into 96 total classes.

---

📋 Additional Information

• load_pickle_code.ipynb demonstrates how the datasets were preprocessed and converted into features.
• Task Process/ folder includes all exploratory code used during development.
• Optimized Models/ folder contains finalized, optimized code for evaluation.

---

👨‍💻 Team

Team Name: MLB_EWHA

• Team Members:
  • Juyoung Yoon
  • Yoonjae Oh
  • Sunghyun Shin
  • Eunwoo Choi
  • Yusun Choi

For any questions or clarifications, please contact us via email