Market Basket Affinity, cannot allocate vector of size 3.3GB error

[Roche]

Hi everyone, 

I have been running many flows using the MB Affinity tool and this is the first time I get this error.  Have done some research on it but do not find much on the MB Affinity tool errors. Have someone had this error before who can help me sort it out?  

I will not be able to share the flow or the data.  But I can give information on the data / flow. There is 405000 records that is processed by the MB affinity tool.  The settings is as follows:

I have also had the number of records per data chunk at 100 000 and at the max which is 256 000.  

Thank you for your help and time!

Rouche

[TheOC]

hey @Roche 

I've seen this error with other R based tools within Alteryx, and it directly relates to how much RAM you have available within your Alteryx Designer. If possible, try to increase the amount of Ram within the runtime settings:

You may also benefit from ensuring your data is the correct types, and optimised, as suggested here:
https://community.alteryx.com/t5/Alteryx-Designer-Discussions/Predictive-Analytics-Linear-Regression...

There is more on this here, too:
https://community.alteryx.com/t5/Alteryx-Designer-Discussions/Cluster-Analysis-Customer-R-tool-Error...

Where it is suggested using another tool, sending data in through chunks rather than all at once. 


Hope this helps,
TheOC

[Roche]

Good morning

Thank you for helping.  I tried that, but it did not produce an outcome with the setting at 50GB.  I will also see what happens when I use half of the data instead.  Hope this one would be a success. 

Have a good week,

Rouche

[leozhang2work]

This reminds me the alteryx challenge 

https://community.alteryx.com/t5/Weekly-Challenge/Challenge-131-Think-Like-a-CSE-The-R-Error-Message...

Assume all the configuration is correct (i.e. data type, column relations), then you will probably need to batch it to a smaller size.

As you mentioned you tried 100,000, then you could also think of dropping some columns if they are not really needed.