Emails are used in professional and personal level as a way of communication. With the passage of time emails are used for advertisement, spreading virus and fraud email for plaguing users of the internet. These type of unsolicited emails are categorized as spam and other legitimated emails are categorized as ham. Over the year several machine learning algorithms are used to predict emails category. In this paper we reflect on the classifier which is good for text classification. We evaluate machine learning algorithm on spam emails detection and outcomes shows naï ve Bayes algorithm gives effective accuracy and precision using WEKA and our email management system which utilize php-ml library hosted by GitHub. Also comparative study with SVM and previous existing system in terms of accuracy and dataset used.