Analysis of Taxi Drivers’ Behaviors Within a Battle Between Two Taxi Apps

This article revolves around a battle between two taxi giants in China, namely Didi and Kuaidad, who are backed by Internet giants Tencent and Alipay themselves. Taxi hiring is the latest thing to go online and is a booming topic for us data analysis enthusiast. We will try to look into what happened in 2014 which changed the Chinese Taxi hiring patterns for good. Data of around 9000 taxis was collected for 37 days and analysed. 

Taxi booking via mobile booking is the new and supposedly easy way of booking taxisaround the world. However what happened in China in 2014, made people study the affects and reasons on what are the possible advantages and drawbacks of this system.For starters, not everyone has a smartphone to book taxi, moreover these facilities are not available in the remotest part of a city. Mobile based taxi booking has its advantages as well. In retrospect of a passenger, who relies on taxi to travel, it can get very difficult to find a taxi in peak rush hours. When you get a taxi in just one click of your smartphone, you don't stand in scorching sun to wait for a taxi to see your waving hand and take you to desired destination. In driver's perspective this is a debatable issue. While some might argue that, it increases driver's efficiency, some may argue that it reduces driver's idel time and others argue that it makes a driver morally corrupt as he prefers to take smaller distance passenger and may bring unfairness to all the users. Next we see the Data analysis done and various  computations done to analyse the situation.

The data set used in this paper includes about 8.3 million taxi trips made by over 9000 taxis during 40 days.

The histogram plot of the numbers of trips made by every vehicle. In the first time period, the battle had not occurred) and in the second time period, the battle is white-hot.
We take a look at the  four taxi service indices: 
1) The distributions of the number of trips served by every vehicle per day during two periods, respectively.
2) The distributions of the idle time lengths of every vehicle before per trip during two periods, respectively.
3) The distributions of the traveling distances per trip during two periods, respectively.
4) The spatial distributions of the origins and destinations of passengers during two periods, respectively.

The graphs below show the recorded data:




The average numbers of passengers served per half hour during two time periods.






The empirical distributions of the idle time lengths of every vehicle before per tr-ip in each day of two time periods.











The empirical distributions of the traveling distance of every vehicle per trip in each day of two time periods.
It was observed that there was a significant change in hiring pattern for short distance trips whereas the pattern remained almost same for long distance trips, say more than 6km. This was because of the promotion money that was kicking in, which was dependent on number of trips made. These short-distance trips did not consume much time and were usually finished one by one so that their time gaps are small.





We further compare the numbers of taxis service happened at different locations with and without money promotion. We grid the central part of Beijing city into 1600 area grids. Each grid covers a 1 km × 1 km square. Above figure shows that the numbers of the incremental trips in 70% of these area grids are nearly zero. In other words, the taxis services are approximately the same under money promotion in such grids.

To conclude, this study was induced by the competition between internet giants who wanted to settle their business in mobile business industry. They started giving promotion money to taxi driver on completion of a taxi trip, which led to some hypothesis about drivers, which include foul practices on drivers side like prefering short distance trips, increased efficiency of drivers etc. Government was forced to shut down these mobile based taxi booking company in peak hours to provide equal treatment to all who seek taxis. 

This is all for this article, more to follow as I read follow up papers.