Learn about the data requirements for user behavior tables before importing them into Quick Audience.
Because the MaxCompute data source type for storing raw data is MaxCompute, and the Analysis Source type used by Quick Audience is ADB 3.0 or Hologres, Quick Audience does not support the following MaxCompute data types to ensure smooth data flow between the computing and analysis sources: BOOLEAN, CHAR, ARRAY, STRUCT, MAP, BINARY. For the full MaxCompute data type list, see 2.0 Data Type Version.
A user behavior table records user actions such as logon, browsing, adding to cart, collecting, and purchasing. The following example shows a MaxCompute wide table:

Field requirements:
|
Field Type |
Field type |
Description |
|
The ID of the user who executes the statement. |
Text type, numeric type |
|
|
The time when the behavior occurred. |
Time |
|
|
behavior channel |
Text type, numeric type |
|
|
The type of the behavior. |
Text |
|
|
Behavior attribute |
Text, numeric, and time |
Examples include products, categories, brands, shops, coupons, discounts, and pre-sale periods. Optional. You can specify multiple fields. |
|
The statistical metric. Valid values: |
Numeric types |
Examples include quantity and amount. Optional. You can specify multiple fields. |
-
All field names are customizable.
-
The table has no primary key. The same user ID can have multiple records. If any value in the behavior time, behavior type, behavior channel, or behavior attribute differs, a separate row is required.
-
Partition tables are supported.
FAQ
Why was the behavior object field removed from the data table sample?
Answer: Starting from V4.4.0, to store reported Event data as a user behavior table, the behavior object field was renamed to behavior attribute. The existing behavior attribute field remains unchanged. You can still use the original user behavior table schema. Behavior object fields in configured schemas are automatically converted to behavior attribute fields. When you perform operations such as User Behavior Table Structure Configuration, Audience Filtering, AIPL Model generation, and Preference Tags, behavior object and behavior attribute configuration items are merged.