jaeW_at_Onyx Coach

I would not get into the habit of JOINING fact tables this will lead to heartache and row growth. if you continue down this path you tend to end up with a ton of single use datasets that only work for the card they power. instead UNION the two tables together (usually I like to differentiate with an Activity Type column.…

@MattLynn and @GrantSmith be careful about that Partition by seller and state. i think you just want to partition by state. check the boundaries between when you switch sellers, you may have unexpected results with that LAG.

@GrantSmith it feels like you're overcomplicating this. if it were me, i'd try writing CASE statements CASE WHEN numerator is not null and denominator is not null then numertor / denominator When denominator is not null then 100 else null END

have you confirmed that your credentials are still valid ?

amendment i would make to @GrantSmith 's ETL. if the employee has not been terminated, then sent the termination date to 1 year from today's date. that way you have a clean FILTER instead of filter on Term Date is NULL. also by setting a rule, you can include it in the data dictionary for how your dataflow / data set…

looooong video about recursive dataflows. in any event. you'll want to have a data column. maybe =today() or whatever the equivalent is in your google sheet. then you have something you can run your comparison against when you build your dataflow. the docs have you do a JOIN. I think I'd recommend doing an APPEND and then…

if you want to calculate livetime value. In ETL you can GROUP BY Client ID and sum(Amount) then JOIN that back onto the transactions. OR you can use a RANK & WINDOW function to calculate cumulative Sum (so you know lifetime value at the time the transaction took place). Or do the same thing in DataSet Views using window…

@GrantSmith , you'd use the CLI to start the execution of the dataset in Domo. presumably, the trigger would be the completion of the Glue job.

if you don't want to use python you can use the Java CLI, https://domohelp.domo.com/hc/en-us/articles/360043437733-Command-Line-Interface-CLI-Tool#scripting or build an app by sniffing API traffic.

Your CSM will be able to ask internally if a custom connector has been built by Domo before. But @MarkSnodgrass is right, he's listed a bunch of options that'll help you get started now.

what question are you trying to answer in your card? there's no reason to group your data in ETL if in analyzer all you want to do is show sales grouped by client, just toss the client on the X-axis and sum(sale total) on the y-axis.

It's a feature switch. Talk to your CSM or email support@domo.com

the problem with window functions in analyzer in this context is that you MUST show date on the axis in order for the lag function to work. I assume you'd only want to show rows if there were 3 consecutive days of xyz status. if my assumption is correct, you'd want to implement your lag calc in ETL so that you have a…

@Ashwin_SG if you use a recursive dataflow to datestamp your ETL based on when you run the ETL, it can be done... but comes with risks you need to consider. If you assume that you ONLY run your recursive once a day then you kind of defeat the purpose of a recursive (which should allow you to run your dataflow as often as…

you could calculate the metric in a dataset view or magicETL before you get into analyzer but then your beast mode won't respond to page filters. alternatively restructure your data. set up your data such that you have one row per user per rubric. this way you can take the SUM of the denoinator instead of trying to do a…

you can't run count(distinct) in a window function and get the 'correct' results. if you imagine it as a sql query. SELECT rubric, count (distinct user_id) FROM Table GROUP BY rubric this would give you expected results. and this is what a beast mode set up with just count(distinct) without the over clause. when you add…

Many Bar charts in Domo support Trellis (search for trellis, https://domohelp.domo.com/hc/en-us/articles/360043439893-March-2019-Release-1) But, Domo doesn't really support an X axis with two metrics. Or at least. I haven't gotten it to work. I feel like it should be possible though.

@fearlessOwl , Domo is a data lake just like S3 or any other flat file storage system. Domo has a database layer, Adrenaline, similar to most massive distributed parallel (jargon-soup :P ) SQL database databases. It does not support no-sql super well. The question isn't 'can Domo work for a use case' it boils down to how…

@ramoshe if you want categorical values on the X and Y axis you'd need a different chart type. Consider a heat map. or pivot table

It sounds like "if there is text then 1. Cleanest logic will be case when text is not null then 1 else 0 end.

oof... this looks like heartache waiting to happen. i usually recommend that you have one dataset per dashboard. so basically take all your datasets and UNION / APPEND them into one dataset called DASH_.. conform your column names as appropriate, and then you know you're applying filters to the right thing. the way you're…

NP. If you're looking for more Domo training or upskilling, I post tutorials and content here: https://datacrew.circle.so/home

@user008875 if you're trying to group orders by number of items ex. (orders with 5 or more items, orders with 2 items , orders with 1 etc), then you definitely can't take the approach you're taking. you need to create a column that counts the number of orders per item. you'll have to materialize it into your dataset (do it…

@cassandra_cook1 what do you mean by "switches to a dataset?" one generally does not simply change the datasets that powers cards. in any case if a filter is set up without specifying the dataset then it will apply that filter to all cards on the page that have datasets that include the column name on which the filter was…

this is for the community slack group. this is for the semi-official Domo slack group.

Value mapper only operates on TEXT. So after you finish you value mapping you'd need to use a tile to Change the Data Type to integer.

oof... don't do it! :P I wouldn't code this into ETL. use the LAG function in analyzer: https://www.youtube.com/watch?v=cnc6gMKZ9R8&t=1s UNLESS you have gaps in your data. in which case you might want to consider the route of creating the 'universe of possible values' as outtlined here:…

you can hack Domo ETLs and make a dataflow APPEND instead of Replace.

cumulative backlog is just a window function sum(sum(1)) over (order by date) the trick is to know if on a date the ticket is still open case when date<= closeDate then 1 else 0 end. course this only works assuming you have a ticket on every single day or you might end up with gaps in your data.…

this can be automated via APIs. use the DatasetAPI or QueryAPI to find out it's age. the tough part will be if you want to retire rows of data (as opposed to entire datasets). you'll need a recursive query with a partitioning model to drop old data. also in the java cli there are functions to assist with setting the…