Analyzing App Data

This page presents an example of app data analysis. In this example, we analyze the transition of a game app's high score.

Analysis example

In this example, we assume that the following KiiObject is used for storing a user's high score.

{ "Score" : 100, "Level" : "Easy" , "AppVersion" : 1 }

  • A key "Score" having an integer value representing the user's score
  • A key "Level" having a string value representing the user's level ("Easy", "Normal", "Hard")
  • A key "AppVersion" having an integer value representing the version of the application (1, 2, 3, ...)

Now, we want to analyze how our users are playing. Based on the above app data, we want to check the following metrics:

  • Metric #1: Average of "Score", sliced by "Level" and "AppVersion".
  • Metric #2: Total count of data (i.e. total score data uploaded), sliced by "Score", "Level" and "AppVersion".

Data example and expected results

Here are some examples of the expected results.

Suppose that a mobile app performs the following KiiObject operations in four days, starting from January 1st.

The metric #1 (the average of "Score") and its aggregated results sliced by the dimension "Level" and "AppVersion" will be as follows:

The score data of ObjectID = 1000 is updated on January 3rd. This update is reflected on the score average graph.

The dimension "Level" is used for calculating the average of "Score" for each of "Easy", "Normal", and "Hard". Similarly, the dimension "AppVersion" is used for calculating the average of "Score" for 1 and 2.

The metric #2 (the total count of data) and its aggregated results sliced by the dimension "Score", "Level", and "AppVersion" will be as follows:

The average score is classified just like the metric #1, but the result by the dimension "Score" looks awkward. This is because the classification is made as a raw value, not as a class value (e.g., 0-9, 10-19, 20-29). The count for each class is thus calculated as one. If we add more data, most of the data will be classified as "Others". See Define the Classes to learn how you can deal with the situation.

How to configure

Here are three steps for analyzing application data (App Data).

  1. Define a metric
  2. Analyze results in the developer portal