Large quantities of data , lack of research and mobile traffic are few of the challenges twitter researchers are facing.
She first did experiments to gather some days. Holdback methodology – where split groups into 2 and collected data on observations. Picked up logs and extracted and processed data.
Hadoop, pig and scalding and R / python at the tools she chose to use for obvious capabilities of each.
She then collected processed data and applied fishers exact count to price of disprove her null hypothesis that the control and exposed groups have equal tendencies.
Having done that she went on to see what is the trend between number of exposure and people actually visiting Vegas. Then she trained her model to understand how much money needs toy be spent to cause a certain amount of result. L
I liked the way she broke down a problem that she gave into an cutting project with lots of insights.