According to research[Link] on suicide-watch dataset[Link], five models are tested on dataset. These models are as follow:
- Naive Bayes (Voting Classifier)
- Random Forest
- Decision Tree
- Gradient Boosting
- XG Boost
At the end, Naive Bayes is choosen as best models on this dataset. As you know, in many psychological counseling institutions and clinics, the information recorded in patient chats and interviews is confidential and cannot be made available to the public. Therefore, the use of cryptographic protocols in analyzing and interpreting this information is of great importance.
In this notebook I impliment a Federated Learning-Based Naive Bayes Model for Detecting Suicide Decision on suicide-watch dataset.