Clusters with Content Type and Content Genre Classification

This user classification will be generated by applying a numerical index to each content type or genre. This index is obtained by averaging the values of each consumption activity variable. Additionally we will add some user profiling features not related to consumption.

If at least 50% of the users universe does not have a total playback time (hours) value greater than zero, the number of playbacks will be used as the baseline variable.

The users without any content consumption within measured period will also be quantified in an additional cluster labeled as inactive users.

Content Type and Genre clusters will be sorted by best consumption variable values criteria.

The clusters will be described and sorted according to the consumption value of the genre or content type that distinguish them. The clusters distinguished with the top values, will be placed first. Higher value clusters will be placed right after. Those that stand out by the lower values will be placed after the higher ones. And finally there will be placed those that stand out for the lowest values.

Content Type Clusters

A total of 4 clusters will be depicted.
The variables used in this classification are the number of content playbacks, the number of hours of content watched or the number of different contents watched. The tenure could be used to determine a differentiation of users who have similar consumption behavior by adding the lifetime bias.

The inactive users cluster will not be depicted for this purpose.

By Content Type / by Content Genre
Audience clustering obtained by averaging the values of content consumption activity by content type or by content genre. This classification is obtained using a relative scale of consumption among the identified clusters.

Top → The highest consumption value of that genre or content type. A cluster in which its users are characterized by the highest consumption of that genre or content type.
Only one cluster has the highest value.
High → Very high relative consumption value. A cluster in which its users are characterized by very high consumption of that genre or content type, but not the highest.
Low → Very low relative consumption value. A cluster in which its users are characterized by very low consumption of that genre or content type, but not the lowest.
Minimum → The lowest consumption value of that genre or content type. A cluster in which its users are characterized by the lowest consumption of that genre or content type.
Only one cluster has the lowest value.

Note:
These clusters are sorted in descending order by the number of users.

Engagement

Engagement