Aim: To make an Innovative Spam Prediction of spam emails using Machine learning modeling techniques and to evaluate their performance. Materials and Methods: The initiative's main goal is to collect samples from two different groups. The K-Nearest Neighbor Algorithm is responsible for Group-1, whereas the Multinomial Naive Bayes Algorithm is responsible for Group-2. For both Algorithms, the same sample sizes were used. 80% of the G-Power will be used in the test set. Result: Data is trained in the given model so that Machine learning can function effectively. Emails are used as inputs for the Multinomial Naive Bayes algorithm, which gives us a probabilistic index of the email and determines if it is spam or not. The K-Nearest Neighbor Algorithm outperforms the Multinomial Naive Bayes Algorithm, and our hypothesis is significant with a significance value of 0.011. Conclusion:  These results were achieved through machine learning models such as Multinomial Naive Bayes, and K-Nearest Neighbors. In this paper,  have demonstrated that for the spam filtering method the most efficient algorithms are KNN and MNB given as they have the highest level of accuracy.