The accuracy of each algorithm used in the study for detecting phishing attempts in blockchain transactions is as follows:

1. Decision Tree Using Gini Index:

   • Achieved an accuracy of 97.33%.

2. Decision Tree Using Entropy:

   • Resulted in an accuracy of 97.22%.

3. Random Forest Classifier:

   • Provided the best accuracy rate of 98.5%.

4. XGBoost Classifier:

   • Attained an accuracy rate of 98.33%.

5. Artificial Neural Network:

   • Achieved an accuracy rate of 88.8%.

The study clearly demonstrates that the random forest classifier and the XGBoost classifier presented the highest accuracy rates among all the models tested for predicting phishing attempts in the blockchain environment.

The dataset used in the study was an open-source dataset from Kaggle containing Ethereum transactional data. It's important to note that the dataset was imbalanced, with a larger number of safe attempts compared to phishing attempts. This imbalance may have influenced the performance and accuracy of the machine-learning models used for phishing detection.

The study focused on detecting phishing attempts in blockchain transactions using machine models. It utilized an open-source dataset from Kaggle containing Ethereum transactional data, which was found to be imbalanced, with a larger number safe attempts compared to phishing attempts. The algorithms used included decision tree classifiers using Gini index and entropy, with accuracies of 97.33 and 97.22% respectively. Moreover, the random forest classifier and XGBoost classifier demonstrated the highest accuracies at 98.5% and 98.33%, outperforming the artificial neural network at 88.8%. The study's findings highlight the significance of accurate algorithm selection and balance for successful phishing detection in blockchain networks.