Alteryx Designer Desktop Discussions

Find answers, ask questions, and share expertise about Alteryx Designer Desktop and Intelligence Suite.
SOLVED

AB trend indicating errors - date, dimension, identifier related

Jaspreet
6 - Meteoroid

Hi,

 

I am getting these errors, despite having my date field configured properly, setting start date to coincide with date fields and giving the right # of periods. I have also imputed any null values with zero, and there is no row without an identifier. Not sure where i am going wrong.

 

AB Trend (7) AB Trend: The range of the dates in the data is not consistent with the test start date and/or the number of periods to base the trend measure on and/or the reporting period of the data.
AB Trend (7) Tool #225: Error in svd(x, nu = 0) : a dimension is zero
AB Trend (7) Tool #225: Execution halted

AB Trend (7) Tool #225: The R.exe exit code (1) indicated an error.
AB Trend (7) Tool #227: The field "Identifier" is not contained in the record.

 

 

4 REPLIES 4
ivoller
12 - Quasar

Can you share the workflow you have so far?

Jaspreet
6 - Meteoroid

Here is the workflow

ivoller
12 - Quasar

A couple of things I noticed.

 

Missing data for 2017-08-01. This will cause problems. I got around it by duplicating data from 2017-07-01 but I don't know what is appropriate in your case.

 

Many accounts had 0 values for all relevant time periods. this causes divide by zero errors in the macro. I resolved by simply removing the accounts with 0 values. This is obviously not appropriate in the real world and you'll need to give some thought to this issue.

 

Make sure the start date is set such that the supporting data includes a full year plus 2 of the periods. I tested against 2017-12-01

 

Set the Show All Macro Messages option in configuration - runtime to see all the errors.

 

See attached

 

Iain

 

Jaspreet
6 - Meteoroid

This is big help, Iain. Thanks!

 

Removing zeros or rows will change the meaning of my analysis though. I want to do some sort of cohort analysis where new subscribers start, old ones leave and some just continue. 

Labels