During any Hurricane, for the possible affected region, usually mandatory and voluntary evacuation orders are called to move the residents of that particular area to a safer zone so that the aftermath of the hurricane can be mitigated. It is useful to know this pattern so that this information can be used during the next evacuation time. However, the collection process of this information is difficult for government agencies.

In this project, R scripts are developed to mine the available tweets data which were tweeted one month before two category five Hurricane Irma, and Maria were originated, and then analyzed those tweets geocoordinates information to find out the county overlap of those tweets. From this county overlap, we can identify the tweet user ID of the possible residents of that hurricane zone. Then using those unique user ids, tweets tweeted during and fifteen to twenty days after the hurricane can be filtered out, and the coordinates of the tweets can be plotted as a heatmap to compare and investigate the overall evacuation pattern of the residents of the affected zone.