目录
此内容是否有帮助?

# Historical Versions of Tags

# I.The Significance of Historical Versions of Tags

Tag values can depict user characteristics, but they can vary by date, such as when an unpaid user first becomes a small R and then a large R. When analyzing, it is possible to draw incorrect conclusions only using the user's current tag values, so try to restore the user situation at the time of the event, which is the meaning of the tag historical version.

Tag historical version is the concept of the latest version of the tag. The tag has at most one historical version under each date, which means the tag value calculated under the tag logic at that time using that date as the tag reference update date.

If you have other tag-related questions, you can go to the User Tags section to view.

# II. Creation of Historical Versions of Tags

# 2.1 Automatic Backup

If automatic backup is selected for tag backup, when the tag is automatically updated, the updated data is written to the historical version of the previous day, i.e. the automatic update task of that day accomplishes two things at the same time.

# 2.2 Manual Backup

In addition to automatic backup, it can also actively calculate the historical version of the specified date, which is mainly used to supplement missing versions, recalculate failed tasks or update label calculation logic and other scenarios.

  • Recalculation: For a date that has been calculated (backed up), you can click the 'Recalculation' button to calculate the data of that date.
  • Batch Calculation: Click the 'Batch Calculation' button to select all historical versions or abnormal versions within the date range for calculation.

Note: Batch calculation of historical versions may take a long time, please choose an appropriate time to perform.

TA system will restore the data at that time as much as possible when manually backing up. If user features are used in the label rules (including virtual event attributes or dimension table attributes using user features), you need to select options in the dialog box:

  • Whether to use user history data: When using user attributes, whether to use the corresponding date user attributes to backup data.
  • How to deal with missing user history data: This option can only be selected when 'Whether to use user history data' is checked. If the user history data of that day is missing, whether to use the user's current data instead or the relatively recent data instead.

# III. Usage of Historical Version of Tags

# 3.1 Use in Analytical Models

If the tag used has a version history, you can specify the desired version when the filter or group item is used.

  • Latest version: calculated according to the user's latest tag value
  • Dynamic matching: calculated by the user's tag value on the day of the event
  • Historical version: based on the user's tag value on a specific date, only the historical version (date) with data can be selected

Note: At present, event analysis, retention analysis, funnel analysis, distribution analysis, path analysis, and interval analysis support tag selection versions, and attribute analysis can only use the latest version of the tag.

# 3.2 Use in Virtual Properties

If the tag used has a historical version, you can specify the needed version when the virtual attribute is created.

  • Latest version: calculated according to the user's latest tag value
  • Dynamic matching: calculated by the user's tag value on the day of the event
  • Historical version: based on the user's tag value on a specific date, only the historical version (date) with data can be selected

Note: Currently, the virtual event attribute supports tag selection version, and the virtual user feature can only use the latest version of the tag.

# IV. Best Applications

# 4.1 View Daily Return User Performance

Through the "Custom Conditions" tab, you can filter out the returned users of the day according to the following conditions:

  • Login today
  • No arbitrary events in the past 7 days (i.e. not login)

Combined with the history tag, you can view the number of returned users per day in history, and you can use it as a filter to see the performance of returned users in the analysis model.

# 4.2 Restore the User's Current Payment Level

User was a small R yesterday and today's payment became a large R. When grouping according to user's payment level, the behavior that the user generated yesterday should be considered as that of small R and today should be considered as that of large R.

After creating the payment level tag, you can view the daily data in the way of 'dynamic matching' in the model group items to get the desired results.