DATA MINING and MACHINE LEARNING. PREDICTIVE TECHNIQUES: ENSEMBLE METHODS, BOOSTING, BAGGING, RANDOM FOREST, DECISION TREES and REGRESSION TREES.: Examples with MATLAB
OBJECTIVE • In number of features (N)-dimensional space, SVM is used to plot a hyperplane that forms a boundary between data points. The hinge function is one of the best loss functions for optimising the margin value. In the following scenario, which is best for linearly separable data, linear SVM is employed. If our model correctly predicts the class of a data point, there is no misclassification. • A random forest is made up of several independent decision trees that each forecast a class forgiven query point, with the ultimate result being the class with the most votes. METHODOLOGY Support Vector Machine algorithm: • In comparison to other classifiers such as logistic regression and decision trees, SVM has a very high accuracy. It is well-known for its nonlinear input space kernel technique. Face identification, intrusion detection, email classification, news article and web page categorization, gene classification, and handwriting recognition are only few of the uses. • SVM is a fascinating algorithm with straightforward concepts. The classifier uses the hyperplane with the most margin to separate data points. PROGRESS • This research solves cyberbullying detection as a binary classification problem, in which we detect two key types of cyberbullying: hate speech on Twitter and personal attacks on Wikipedia, and classify them as containing or not containing cyberbullying. • Support Vector Machine (SVM) for Twitter Hate Speech and Random Forest Classifier for Personal Attacks are used in the suggested system. • In number of features (N)-dimensional space, SVM is used to plot a hyperplane that forms a boundary between data points. The hinge function is one of the best loss functions for optimising the margin value. In the following scenario, which is best for linearly separable data, linear SVM is employed. If our model correctly predicts the class of a data point, there is no misclassification. CHALLENGES FACED 1.Less Accuracy 2.Existing system strategies just look for patterns in the data that already exist. 3.The majority of current system procedures are manual processes that rely on human intervention and decision- making. FUTURE PLAN • Incorporate advanced deep-learning models: Enhance the performance of detection systems. • Develop new groups of features: Capture textual similarities and repetitive time aspects in comments. • Use multiple deep learning methods: Improve detection accuracy.
DATA MINING and MACHINE LEARNING. PREDICTIVE TECHNIQUES: ENSEMBLE METHODS, BOOSTING, BAGGING, RANDOM FOREST, DECISION TREES and REGRESSION TREES.: Examples with MATLAB