Valued event analysis

Motivation

• more then 100 events in Umeng, but all of them are valued? or wasted?

Try to find some relevance

• data source : Umeng_Event_GetUniqueUsers.csv (daily app unique users)

• feature : average, maximun and minimun of unique users on different period

• K-means method: k-means clustering is a method of vector quantization, originally from signal processing, that is popular for cluster analysis in data mining.

Find their group

• ios_cluster

• group1(valued) means that events now these events have many unique users:

• group5&6(wasted) means that events less to use or maybe that issues are seldom needed:

• android_cluster

• android device can use the same method to separate different group.

• group1(valued) means that events now these events have many unique users:

the different between two devices is App_First_Launch(android) and Launch_Soccer(ios).

• group5&6(wasted) means that events less to use or maybe that issues are seldom needed:

• group6 event don’t have any data(maybe they haven’t done)

• the explains in these two methods are 94.71%(ios) and 93.71%(android).

• finished data