Performance comparison of mushroom type classification based on multi-scenario dataset using decision tree c4.5 and c5.0

Indonesia has a tropical climate that supports mushroom growth. Mushroom classification into poisonous and non-poisonous mushrooms. Identification of the type of mushroom is vital because mushrooms, especially poisonous mushrooms, risk causing potential hazards to humans, such as causing serious illness and even death. This study aimed to identify the fungus type using a computational approach, namely the Decision Tree C4.5 and C5.0 Algorithms. This research contributes to using multi-scenario datasets and comparing the performance of the C4.5 and C5.0 decision tree algorithms. The dataset used is a fungal classification dataset obtained from kaggle.com. The method stages in this research are literature study, data collection, and data preprocessing, which includes a data cleaning process and a partitioning process for multi-scenario datasets. Afterwards, the Decision Tree Algorithms C4.5 and C5.0 were implemented using the sci-kit-learn library. The last step is to do a performance comparison using the confusion matrix. The results showed that identifying poisonous mushrooms using the Decision Tree C5.0 Algorithm obtained an accuracy of 97.05% for scenario 1, 97.00% for scenario 2, and 97.11% for scenario 3. At the same time, the Decision Tre C4.5 algorithm yielded an accuracy. by 96.92% for scenario 1, 96.90% for scenario 2, and 97.05% for scenario 3. Based on the comparison of the performance of the classification results, we conclude that the Decision Tree C5.0 algorithm in scenario 3 has the highest accuracy for fungal identification poisonous.