A Comparative Study of Machine Learning Algorithms for Mushroom Toxicity Prediction
Students & Supervisors
Student Authors
Supervisors
Abstract
Predicting mushroom toxicity is a huge issue in food safety, and machine learning (ML) is a potentially helpful approach which allows for the automation of classifying edible and poisonous mushroom species. This paper will report on the application of a variety of machine learning models (Decision Trees, Random Forest, K-Nearest Neighbors (KNN), and Gradient-Boosted Trees (XGBoost)) to predict the edibility of mushrooms based on a large dataset with a mixture of features. It contains numeric variables (e.g., cap-diameter, stem-height, stem-width), and categorical variables (e.g., cap-shape, cap-color, odor), and the target variable is binary (edible = 1, poisonous = 0). The methodology includes a thorough preprocessing of the data, which is consistent and eliminates information leakage by using a Column Transformer to encode and normalize features. The data is separated into a training (75%) and testing (25%) data sample, and models are evaluated in terms of accuracy and ROC-AUC scores. The threshold optimization is performed to achieve maximum classification performance, with minimal emphasis on misclassifying poisonous mushrooms. Moreover, there are other feature engineering methods, such as interaction construction and feature importance ranking, which provide insight into the model's decision-making process. Learning curves, confusion matrices, and feature importance plots are the visualizations of the results. The visualizations of the results are learning curves, confusion matrices, and feature importance plots. The paper will conclude by reviewing the deployment perspective, where model retraining and drift monitoring are particularly relevant in a real-world setting. The approach provides strength and understanding, which is critical in the application of machine learning models in food safety.
Keywords
Publication Details
- Type of Publication:
- Conference Name: 7th International Conference on Integrated Sciences (ICIS) 2025
- Date of Conference: 25/10/2025 - 25/10/2025
- Venue: Eastern University Campus, Ashulia Model Town, Dhaka, Bangladesh
- Organizer: Eastern University, Bangladesh