Fake Profile Detection and Stalking Prediction on X using Random Forest and Deep Convolutional Neural Networks

This study employs Random Forest (RF) and Deep Convolutional Neural Networks (DCNN) to predict stalking behavior on X and detect phony profiles. The source of the dataset was Kaggle. The model was developed and evaluated using the Object Oriented Analysis and Design (OOAD) methodology. Utilizing the Python computer language, the RF&DCNN algorithms were implemented. Real-time detection and prediction are provided by the algorithms, which process the input data iteratively and update the model parameters in response to fresh observations. Statuses_count, followers_count, friends_count, favorites_count, and listed_count are among the input parameters provided into the model. By including these parameters in the model, profiles can be predicted effectively and with accuracy. Based on the research, an accuracy level of 93.89% with an error rate of 6.104 was achieved. With an accuracy rate of 86.57% and an error rate of 13.43%, the proposed model outperformed the current one in terms of effectiveness. The outcomes show how well the RF and DCNN based prediction model works to identify fake profiles and predict stalking. By putting out a novel method for identifying phony profiles and forecasting stalking utilizing RF and DCNN, this study advances the field of anomaly detection operations.