Merging rows with same timestamp for time series

[albert]

Hi Grafana Community,

I would like to create a panel to visualize user agent trends over time. The time series should look something like this with a line for each unique user agents.

I have a GraphQL datasource that returns data in the following format:

Timestamp	User Agent	Count
2023-09-10T00:00:00Z	Firefox	200
2023-09-10T00:00:00Z	Safari	300
2023-09-10T00:00:00Z	Chrome	400
2023-09-10T00:01:00Z	Firefox	250
2023-09-10T00:01:00Z	Safari	350
2023-09-10T00:01:00Z	Chrome	450
2023-09-10T00:02:00Z	Firefox	300
2023-09-10T00:02:00Z	Safari	400
2023-09-10T00:02:00Z	Chrome	500

Unfortunately, if I plot this data directly, the time series ends up looking like this:

This is seemingly because the time series visualization expects a row for each timestamp and a column for each series. The data needs to be structured as such:

Timestamp	Firefox	Safari	Chrome
2023-09-10T00:00:00Z	200	300	400
2023-09-10T00:01:00Z	250	350	450
2023-09-10T00:02:00Z	300	400	500

The GraphQL API I use does not allow me control over how the data is presented, so I need Grafana to transform it to a structure that can be displayed in a time series. How can this be achieved?

If it helps, I am using Grafana Cloud and the Infinity datasource plugin to interact with a GraphQL API.

Thanks a lot in advance!

[albert]

I managed to do this with the “grouping to matrix” transformation.

[albert]

It appears a simpler solution is to use the “partition by values” transformation:

[albert]

Seems just selecting “Time Series” format in Infinity datasource settings does the trick too - without transformations