Hi,
We’re using v7.6 and we’re very satisfied to use approx_percentile in our reports.
Now, we have a new generation of reports that require histograms.
Is there an optimized function for histograms ?
Great question Nabil! I’m checking with the team and will update you soon. Thanks for your patience. 
Hi Nabil! We hope you find this reference helpful. Please keep us updated on how it worls out for you! Cheers 
Statistics for a table can be collected and updated on an ad hoc basis by running the ANALYZE command. If autostats are disabled for a table, ANALYZE collects column statistics on every column, as well as histograms on chosen columns. You may designate which columns to collect histograms on with the ANALYZE TABLE table_name COLUMNS … {ENABLE | DISABLE} commands, as described in the docs. Collecting column statistics is strongly recommended for optimal query performance. Collecting histograms is recommended in most situations.
See Histogram vs. Sampling for more details.
@Nabil the way I read your question, you want to compute approximate histograms in a SELECT and return them to the client. We don’t support that. But you could code it in a user-defined aggregate function, especially now that we support Wasm UDFs on SingleStoreDB Cloud. Wasm UDFs can be used to define a UDAF, which is just a package of several UDFs.
1 Like