In the modern days of life, people use many social media sites for information sharing among friends, relatives, and others for personal, business, and official purposes. The use of social media platforms is also raising serious issues in the form of cyberstalking. Cyberstalking has been identified as a growing anti-social problem that affects educational institutions, victims, and entire human society. An intelligent system is required to detect cyberstalking in social media. In this paper, we proposed a cyberstalking detection model and analyzed the performance of six popular supervised machine learning algorithms, namely Logistic Regression, Support Vector Machines (SVM), Random Forest, Decision Trees, K-Nearest Neighbor, and Naive Bayes. These machine learning algorithms were implemented with two feature extraction methods, Bag of Words and TF-IDF, on two datasets of different sizes and distribution containing 35734 and 70019 comments and tweets, respectively. Performance of algorithms was measured in terms of Accuracy, Precision, Recall, f-score, training time, and prediction time. Our experimental results show that Logistic Regression and Support Vector Machine were top performer algorithms for both datasets with both feature extraction methods. Logistic Regression (92.6% with BOW and 92% with TF-IDF) and Support Vector Machine (92.5% with TF-IDF and 91.9% with BOW) achieved the highest accuracy on dataset-1. Logistic Regression and Support Vector Machine also achieved the highest Precision (96.4% and 96.6% respectively) and F-Score (94.3% and 93.8% respectively), while Naïve Bayes provides the best Recall (97.6% with TF-IDF on dataset-1) for both datasets.