ML_MiniProj

Essay Classification Project

Overview

The Essay Classification Project is a machine learning endeavor designed to categorize essays into specific genres or prompts. The project employs feature extraction techniques and two distinct models—XGBoost and AdaBoost with a logistic regression base estimator—to achieve accurate essay classification.

Key Components

• Feature Extraction: The feature_extraction.py module utilizes various functions to extract essential features from essay texts. These features include sentence and word counts, punctuation presence, and the identification of specific words or phrases.

• Model Training: The model_training.py module is responsible for training machine learning models. It employs XGBoost and AdaBoost with a logistic regression base estimator. The ROC AUC scores are evaluated and presented in a bar graph for model performance comparison.

• Data Handling: The data/ directory stores input data, exemplified by train_essays_7_prompts.csv. The output/ directory contains the generated submission file, submission.csv, which includes the average predictions of both the XGBoost and AdaBoost models.

Usage

1. Install the required dependencies:

   pip install -r requirements.txt
   

References

This is an Implementation of the conclusions from the paper cited below

• Author(s) Heather Desaire,Aleesa E. Chua,Min-Gyu Kim,David Hua . "Accurately detecting AI text when ChatGPT is told to write like a chemist." Elsevier, 2023, DOI or Link.