Video user behavior and recommendation system evaluation KPI-part

 

 

Problem	KPI
Is the number and ratio of users in the recommendation area significantly increased?	Compare the number of users in the recommendation area and their proportions with those before.
Are there significant differences in the use of recommendations for new and old users?	Percentage of New and Old user recommendation zones in use by category, percentage of VV generated by new and old user recommendation zones by category
Does the VV proportion produced by the recommendation area significantly increase in the total VV proportion?	Compare the percentage of VV in the recommendation area to that before the recommendation system goes online.
Does the recommendation area significantly increase the VV volume?	Compare the VV volume and percentage in the recommendation area with those in the same region before the recommendation system goes online.
Field	Description
ID	Unique access ID
UID	User ID
Region ID
Access time
PV count
User Type	0: new user, 1: Old user, 2: Unknown
VV Access Mode	0: access to the theater, but no VV is generated. 1: VV is generated, but VV is not continuous. That is, VV is not generated in the recommendation area. 2: VV is generated, VV is generated in the recommendation area.
VV times	VV times per access cycle
In VV times, refer indicates the number of playbacks on the theater.
Refer channel of the theater's first VV
Access duration
User Access URL