Train ML Model Option, Which one is better?

[Tejaskumar_Darji]

Hello,

For the ML Model which option should be more appropriate and will yield better results?

------------------------------
Tejaskumar Darji
------------------------------

[PatrickAucoin]

I believe that the automatic setting is 1,000 documents, but you can set it to any value between 50 and 5,000. This setting should depend on the current accuracy of the model. The tradeoff is more frequent feedback iterations versus better processing efficiency / performance.

In general, we recommend to lower this setting when you are starting to train a new model, as fewer documents have been observed and accuracy is expected to be lower. Once accuracy is at an acceptable level, you can increase the setting.

tl;dr version: Lower it when accuracy needs improvement, raise it when accuracy is fine.

------------------------------
Patrick Aucoin
Senior Product Consultant
Blue Prism
------------------------------