Distribution Analysis
Distribution Analysis aggregates the selected analysis entity by the number of times or days it completes events, or by event properties, and divides the results into ranges so that you can see the number and proportion of users in each range.
With Distribution Analysis, you can answer questions like these:
- What were the proportions of users in different login-day ranges last month?
- Since the server launched, how many users have a cumulative payment amount of more than 1000 yuan?
- Are the acquisition and consumption of a virtual currency resource healthy? Is there a large stockpile of currency that needs to be consumed in time?
Configure the analysis entity and events
The Distribution Analysis model aggregates data by the selected analysis entity and counts how each entity participates in events. You can count by times, days, or hours, or by a calculation method based on an event property. You can also use a formula as the counting criterion.
For example, you can count the number of days users logged in last week and divide them into ranges, such as treating users who logged in on 4 or more days last week as highly active users, and then view the number and proportion of these users. Note that only those who have triggered the selected event are included in the statistics. If a user didn't log in last week, the user isn't placed in any range.
As another example, to analyze each user's diamond balance and determine whether the current production and consumption of virtual currency are balanced, you can set the formula "amount obtained in Get Diamonds - amount consumed in Consume Diamonds" to get the diamond balance, divide the balances into ranges, and view their proportions.
To further analyze the users in a specific range, turn on Also Show. On top of the ranges, Distribution Analysis then calculates the Also Show metrics (that is, the Events Analysis metrics of these users) based on the event data of the users in each range.
For example, you can first divide users into low, medium, and high tiers by login days, and configure the per-user revenue as an Also Show metric to quickly compare the payment of users with different activity levels.
As with other models, you can also filter data or compare with group by.
Visualize the distribution with charts
By default, Distribution Analysis shows detailed data in a table. In the table, you can view the number of entities, the proportion, and the Also Show metric data for all users and for each range. You can also click through to the User List or save the users as a result cohort. Hover over a cell to see the detailed statistical criteria of the data.
To view the number of users in each range, you can select one of the following chart types:
- Stacked Line
Stacked Line shows how the number of users in each range changes over time.
- Percentage Line
Percentage Line shows how the proportion of users in each range changes over time.
- Histogram
When the time granularity you select is Total, you can view the number of entities in each range within the time range in the histogram.
If you want to View [Also Show] Only, you can select one of the following chart types:
- Line
The line chart shows how the Also Show metrics of entities in each range change over time. It's available when the time granularity isn't Total.
- Bar chart
When the time granularity you select is Total, you can view the Also Show metric data of entities in each range within the time range in the bar chart.
When you use Group by, all chart types except the table can show data for only one group value. If you've added group-by items, click Groups to switch between them. By default, the data for the total (without group-by items) is shown.