As we strongly believe that interesting and significant relationships can be revealed by simultaneously considering seasonal information, people from different districts, usage of mobile applications and so on, we aim at showing how to combine various information and prototyping a tool for analyzing the data.

To this end, we first visualize through heat maps how people used the mobile phone with respective to the geographic information and time period spanning three months. We find remarkable peaks of mobile usage around golf clubs during Thanksgiving, and weakened signal strength around shopping malls during that time. This observation understandably shows that some people took use of the Thanksgiving holidays to relax at golf court, where the mobile signal was strong; while some others poured into shopping malls on the Black Friday, where the signal was weakened due to signal overload.

Moreover, we visualize how the usage of mobile applications distribute geographically at the app category level. We strongly believe that, by using census data, we expect to see people will favor different types of app's geographically, as people from wealthy communities will care much about the their health conditions thus app's related to health might be largely used in that area.

As a team, we four member work on different parts, including data cleaning, auxiliary data collection, visualization, data mining, etc. While data provided are not clean in the sense that a lot of data are mis-placed in the csv table, data cleaning is an important step. Moreover, while diving into deeper analysis, how to efficiently process data is another big problem. To alleviate this, we widely use hash table structure. Just name a few.

Based on what we've built, we are proud of what we've discovered at unique angles, yet we expect more discoveries if given more time. As there are numerous aspect to explore in the future, we would continue to make use of geographic information to analyze the trends of mobile app usage. We believe this will give valuable predictions to companies how to improve their products timely, and how/when/where/to whom to place advertisement.

Share this project: