Aggregate a Beast Mode
Hello,
I am working with a retail dataset and need to calculate beginning and ending inventories based on a user selected date range. The calculation is straight forward but the problem is aggregating the result so it displays on one row per item. The goal is a table card of items reporting beginning inventory, ending inventory, units, and dollars sold in the selected period.
Example calculation for beginning inventory:
case
when Date
=min(Date
) OVER (partition BY Store
,Item No
,Variant No
) then sum(NAV_OnHand
)
else 0
end
All attempts to aggregate the case statement fail. The problem is Domo displays a row for each date in the selected period and I know this is due to the evaluation of the date on each row.
Any ideas how I can keep this calculation dynamic based on user date selection. The only alternative I can think of would be to have a row for every item and every day with the beginning and ending inventory calculated in the dataflow. I would prefer not to do that for 100s of thousands of items unless there is no other way to provide this functionality.
Thank you,
Chris
Chris
** Was this post helpful? Click Agree 😀, Like 👍️, or Awesome ❤️ below **
** Did this solve your problem? Accept it as a solution! ✔️**
Best Answer
-
@Chris_Wolman without knowing what your data really looks like is hard for me to picture a full path to follow. In general if what you want is to be able to know what was the inventory at any given point in time (captured using a Variable as @MichelleH suggested), is to have your Raw data being the movements to inventory by date, so you'll have entries whenever things are sold and whenever you acquire new things to sell, and then doing a best mode that aggregates those for any date prior or on the provided variable:
SUM(CASE WHEN
movement date <=
Variable date
ANDmovement type
= 'IN' THENqty
END) - SUM(CASE WHENmovement date <=
Variable date
ANDmovement type
= 'OUT' THENqty
END)If you have 2 date variables for the Start and End you'd simply have two Beast Modes that tells you the inventory at each point in time, and the advantage of this is that you can also show how many IN movements were performed between the two dates vs how many movements out (I guess your inventory can go up if you acquire multiple units to sell during the period).
So using your example table, you'd need to do an ETL that would transform that to:
ITEM A 4/15/2023 IN 10
ITEM A 5/5/2023 OUT 2
ITEM A 5/6/2023 OUT 1
You can go even more detailed if besides showing the direction of the movement, you also specify why, like if you have OUT because of sales, OUT because of damaged good, IN by purchases or IN by returns, so this model would be easy to scale to the business needs, but you need to figure out first how to get the individual movement list out (ETL would be your best bet if you don't already have a dataset with this information).1
Answers
-
@Chris_Wolman Typically, aggregation functions should be outside of the entire case statement in order to return a single row. However in this case the min date is also causing an issue because you would have an aggregate inside an aggregate (min inside a sum). Based on your use case, I would suggest setting up two date variables for your start and end dates so you can avoid the second aggregation with a beast mode like this:
sum(case when `Date` = `Start Date (Variable)` then `NAV_OnHand` else 0 end)
1 -
Hi @MichelleH ,
Variables will not work in this case because there may not be a row in the data for that start date. Example, user selects a start date of 5/1/2023 and end date of 5/15/2023. The earliest date available for the item is 5/5/2023. I would need the calculation to use the earliest date in this scenario which is why the comparison is against the min(date). If there was a beginning inventory for every item and every date, it would work. I can create the dates in the dataflow but my resulting dataset will be in the hundreds of millions rows, which is I was hoping to keep the calculation dynamic.
Thank you,
Chris
Chris
** Was this post helpful? Click Agree 😀, Like 👍️, or Awesome ❤️ below **
** Did this solve your problem? Accept it as a solution! ✔️**0 -
@Chris_Wolman Can you provide some more information about the business case for why the starting inventory for an item that wasn't available until 5/5 would not be 0 for the period between 5/1 and 5/15?
0 -
Sure, here is an example of the data.
The above illustration shows 10 units received on 4/15/2023. No sales for the item until 5/5/2023. The expected result is one row for Item A with beginning inventory = 10, ending inventory of 7, and sales of 3.
This is why we are using a windowed function to get the beginning inventory. Not every item will have a row for the exact date range selected.
Chris
** Was this post helpful? Click Agree 😀, Like 👍️, or Awesome ❤️ below **
** Did this solve your problem? Accept it as a solution! ✔️**0 -
So on your data, you have the current number of items in a field
OnHand
? If this is the case, this would seem an ideal scenario for a flex table instead, have you looked at this option and configured the columns to show First Value, Last Value and the change?1 -
I agree with @marcel_luthi I think a flex table or some other time series chart would be better for visualizing this.
0 -
Wouldn't the flex table give me the wrong beginning inventory because the 4/15/2023 row would be excluded when filtering 5/1-5/15 on the card?
Chris
** Was this post helpful? Click Agree 😀, Like 👍️, or Awesome ❤️ below **
** Did this solve your problem? Accept it as a solution! ✔️**0 -
@Chris_Wolman The 4/15/2023 row would be excluded with those filters in all of these scenarios. I guess I'm still not clear on what questions you are trying to answer with this data.
0 -
@Chris_Wolman without knowing what your data really looks like is hard for me to picture a full path to follow. In general if what you want is to be able to know what was the inventory at any given point in time (captured using a Variable as @MichelleH suggested), is to have your Raw data being the movements to inventory by date, so you'll have entries whenever things are sold and whenever you acquire new things to sell, and then doing a best mode that aggregates those for any date prior or on the provided variable:
SUM(CASE WHEN
movement date <=
Variable date
ANDmovement type
= 'IN' THENqty
END) - SUM(CASE WHENmovement date <=
Variable date
ANDmovement type
= 'OUT' THENqty
END)If you have 2 date variables for the Start and End you'd simply have two Beast Modes that tells you the inventory at each point in time, and the advantage of this is that you can also show how many IN movements were performed between the two dates vs how many movements out (I guess your inventory can go up if you acquire multiple units to sell during the period).
So using your example table, you'd need to do an ETL that would transform that to:
ITEM A 4/15/2023 IN 10
ITEM A 5/5/2023 OUT 2
ITEM A 5/6/2023 OUT 1
You can go even more detailed if besides showing the direction of the movement, you also specify why, like if you have OUT because of sales, OUT because of damaged good, IN by purchases or IN by returns, so this model would be easy to scale to the business needs, but you need to figure out first how to get the individual movement list out (ETL would be your best bet if you don't already have a dataset with this information).1 -
Thank you @marcel_luthi and @MichelleH . I believe the use of variables for start and end date along with identifying inventory movement is the solution that will allow us to calculate the beginning and ending inventories for the period any user selects.
Chris
** Was this post helpful? Click Agree 😀, Like 👍️, or Awesome ❤️ below **
** Did this solve your problem? Accept it as a solution! ✔️**0
Categories
- All Categories
- 1.8K Product Ideas
- 1.8K Ideas Exchange
- 1.5K Connect
- 1.2K Connectors
- 296 Workbench
- 6 Cloud Amplifier
- 8 Federated
- 2.9K Transform
- 100 SQL DataFlows
- 614 Datasets
- 2.2K Magic ETL
- 3.8K Visualize
- 2.5K Charting
- 729 Beast Mode
- 53 App Studio
- 40 Variables
- 677 Automate
- 173 Apps
- 451 APIs & Domo Developer
- 45 Workflows
- 8 DomoAI
- 34 Predict
- 14 Jupyter Workspaces
- 20 R & Python Tiles
- 394 Distribute
- 113 Domo Everywhere
- 275 Scheduled Reports
- 6 Software Integrations
- 121 Manage
- 118 Governance & Security
- Domo Community Gallery
- 32 Product Releases
- 10 Domo University
- 5.4K Community Forums
- 40 Getting Started
- 30 Community Member Introductions
- 108 Community Announcements
- 4.8K Archive