Mouse Movement Detection: Human vs. Bot Prediction

This repository contains a machine learning project that analyzes mouse movement patterns to predict whether the user is a human or a bot. The project leverages various data processing and machine learning techniques to build an effective prediction model.

Project Overview

The main notebook, MouseMovementDetection.ipynb, performs the following tasks:

1. Data Analysis and Visualization:

   • Exploratory data analysis (EDA) of mouse movement patterns.
   • Visualization of user actions and directions using heatmaps, count plots, and correlation matrices.

2. Feature Engineering:

   • Identification of significant features from raw data.
   • Dimensionality reduction techniques like PCA and T-SNE.

3. Model Training and Evaluation:

   • Implementation of various classification models:

     • Decision Trees
     • Support Vector Machine (SVM)
     • Random Forest
     • AdaBoost
     • Multi-Layer Perceptron (MLP)
     • LightGBM

   • Evaluation using metrics like accuracy, classification reports, and ROC curves.

4. Hyperparameter Tuning:

   • Use of GridSearchCV to fine-tune hyperparameters for models like AdaBoost and Random Forest.

How to Use

• Clone the repository:

  git clone https://github.com/AnnonymousBanda/MouseMovementDetection.git
  cd Mouse-Movement-Detection

• Open MouseMovementDetection.ipynb in Jupyter Notebook or another compatible environment.

• Run the notebook cells sequentially to:

  • Explore the dataset.
  • Train and evaluate the models.
  • Visualize the results.

Project Highlights

• Visualization: Clear and insightful plots to understand mouse movement behavior.
• Machine Learning Models: Comparison of multiple classifiers to identify the best-performing model.
• Interpretability: Focus on understanding which features are most influential in distinguishing between humans and bots.