Hi @D Sai Trailokya ,
he recommended number of records for training a good solution is from 30,000 through 300,000.
If you submit less records than the recommended minimum then solution may not cover all the scenarios which may leads to bad/less accurate predictions.
If you submit more than 300,000 records, the most recent 300,000 records are used to train the solution. Use only authentic records from the database.
Also depending upon the model, Servicenow sets minimum and maximum limit of records needed for training for every model definition.
- Similarity 10 - 100000
- Clustering 100 - 100000
- Classification 10000 - 300000
- Regression 10000 - 300000
These limits are controlled by the system properties - https://[instance].service-now.com/sys_properties_list.do?sysparm_query=nameSTARTSWITHglide.platform_ml.api.max%5EORnameSTARTSWITHglide.platform_ml.api.min&sysparm_view=
So it is recommended to have a training set with required and quality records so that it can cover most of different scenarios which will help in generation of high precession model.
Taken from : https://www.servicenow.com/community/platform-analytics-forum/predictive-intelligence-data-limit/m-p...
