In this research, machine learning methods are used to automatically recognize fake news which is currently a major threat on the web regarding acquired electronically information validity and accuracy. After   that, calculate the performance of both the Logistic Regression (LR) model and the Navie Bayes (NB) classifier to avoid fake news. Comparative analysis suggests that Logistic Regression (LR) model delivers   higher   accuracy than Navie Bayes (NB) classifier in   fake   news   detection. The performance of both classifiers is calculated by confusion matrix, (ROC-AUC) Curve and metric evaluation report for which cross-validation technique used. Accuracy achieved by Logistic Regression (LR) Model is around 98%, overall average accuracy in correctly identifying a fake news compared to Naive Bayes (NB) classifier which could only achieve approximately 93 %, keys matrices such as recall, precision, recall, F1-Score, F2-Score computed on both the models gave values closer to 0.98 at Logistic end better than Navie Bayes (NB) best obtained value of corresponding metrics i.e., 0.93. In addition, the K-fold cross validation is also applied to illustrate our generalized predictions for unknown data. Both models showed reasonable statics in performance. A Logistic Regression (LR) classifier scored better in accuracy than a Navie Bayes (NB) classifier and had a mean prediction of 96% vs 93% for the Navie Bayes (NB) model. Other computed metrics backed this up as well. The large performance gap calculated demonstrates the great utility of logistic model over Naive Bayes (NB) where this can lead to handle the problem under misinformation. For fake news given different thresholds, future research should focus on how well the applied probabilistic calibration responds to real-world application scenarios (ROC-AUC).