Twitter trending topics represent the top conversation happening every day on social media. Although there are some researches about how to classify those hashtags, none of them are connecting the hot topics with the time trend of each topic after they appear in the trending board. The purpose of this project is applying data mining on Twitter data to analyze the time trend of those trending hashtags and use K-means clustering to classify those topics according to their trends. After that, by observing the patterns, anomalies and correlations of data in each cluster, define these clusters and use the results to make a prediction about the trends features of a topic, in terms of the moment when it becomes “unpopular” and when it will regain the attention from the society.