Analysis of automatically splitted/partitioned dataset

Tina_liaddana

Hello,

I have a database of customers (number of transactions for each day) and I want to make an analysis of each customer independently.

I want to automatically split data by customer and then make charts/statistics of each sub-data and finally have them all in the same dashboard (the charts and statistics of all customers).

Is this possible in Dataiku?

For the split part, I know that the recipe "split" and "partitioning" can be helpful, but using "split" (I need to split the data by customers manually, same for partitioning).

Thank you in advance,

Turribeach

In Dataiku partitions are not used in the same way than databases. They basically allow you to avoid having to recalculate older partitions of data when new data comes in. So not really your use case.

Separating the data by customer seems counter intuitive to me. Most business have lots of customers so the solution is not going to scale well. Usually people will load data for all customers and then allow the visualisation tools to filter by customer so you can see. Have a look at the Dashpoard filters: https://doc.dataiku.com/dss/latest/dashboards/filters.html

Alternatevely what you could do is to have a Customer ID variable in your project variables and load data for each customer in a separate scenario run. But the data will be stored in the same output tables.

Tina_liaddana

Hello,

Thank you first for the clarification.

Yes, I do have a customer id variable but this is not really my problem.

My question is rather is it possible to make an analysis (calculate some metrics and specific functions on the data of each customer separately) and then create charts relative to that customer, and finally group all results (charts) in one single dashboard.

n.b: The functions that I want to calculate doesn't figure in the "group by" recipe.