Post-Aggregate report Filters in Time Series Plots

[Percentile95]

Hello SAS Visual Analytics (SAS-VA) users,

I have a simple question regarding time series plots in SAS-VA. In my plot, I have the date (DD-MM-YYYY) on the X-axis and a measure on the Y-axis, consisting of a numeric variable aggregated to a mean value. I would like to exclude those dates where the frequency of the numeric values is below twenty. In other words, I want to exclude days where the averages are calculated from fewer than 20 numeric values.

The problem is that I cannot apply a frequency filter directly to the time series plot. When I set the frequency filter to > 20, no results are returned.

Is there a workaround for this problem, or am I overlooking something?

[BrunoMueller · Accepted solution]

A time series plot can not filter on aggregated values.

However you could mark those points with a different color, see example.

The lower graph uses an aggregated data source, as you now have just a measure you can apply a filter.

Both graphs highlight / filter frequency >= 13

the y axis has been set to be fixed from 0 - 16

[Percentile95]

Dear Bruno Mueller,

Thank you for your response—much appreciated 🙂

I have already attempted to create an aggregated table within SAS-VA to filter when using a time series plot. However, my problem is that I’m working with very large datasets (more than 20 million records), which leads to long processing times or error messages.

That said, I’ve used your idea of coloring data points using display rules, and report users find it agreeable.

Thank you for clarifying that filtering an aggregated measure in Time Series plots is not possible, yet still offering a valuable workaround!

Best regards Percentile95

[BrunoMueller]

Thanks for the feedback.

[BrunoMueller]

One idea, I assume the time series plot is not from 20 million rows. Aggregated datasources support filters where you can also use parameters.

[Percentile95]

Thanks for your support! I’m currently working in the field of Clinical Biochemistry, focusing on lab tests performed on blood samples. My current dataset contains approximately 22 million records. Unfortunately, using aggregated data sources in SAS Visual Analytics (SAS-VA) is challenging due to performance issues; it's either too slow, or error messages occur.

However, users are primarily interested in a subset of the data corresponding to specific blood tests, such as COVID-19 tests or kidney function measurements. By allowing users to specify a single test using a parameter directly on the SAS-VA data source to filter for a single blood test, I can significantly reduce the number of records to well below 20 million. This makes using aggregated data sources feasible and enables efficient filtering in time plot series.

So, I believe I’m on the right track 😊