Describe the issue
When using the calculate_model_statistics function, the generated file dmcas_fitstat.json, in the "Data" section of the json, contains float/integer values for any statistics instead of string values.
Importing this file in SAS Model Manager results in "NaN" values when comparing this model to another.
See the screenshot below.
Wrapping these values in quoted strings corrects the issue.
Attaching the dmcas_fitstat.json file where the training values where manually wrapped as strings (correctly displayed in SAS Model Manager) and the validate/testing ones are originals from the sasctl function.
dmcas_fitstat.json
To Reproduce
Use the function calculate_model_statistics with any dataset and import that file to a model's files section in SAS Model Manager
Expected behavior
The Compare Section in SAS Model Manager should display the model's statistics values, not NaN.
Describe the issue When using the calculate_model_statistics function, the generated file dmcas_fitstat.json, in the "Data" section of the json, contains float/integer values for any statistics instead of string values. Importing this file in SAS Model Manager results in "NaN" values when comparing this model to another. See the screenshot below.
Wrapping these values in quoted strings corrects the issue.
Attaching the dmcas_fitstat.json file where the training values where manually wrapped as strings (correctly displayed in SAS Model Manager) and the validate/testing ones are originals from the sasctl function. dmcas_fitstat.json
To Reproduce Use the function calculate_model_statistics with any dataset and import that file to a model's files section in SAS Model Manager
Expected behavior The Compare Section in SAS Model Manager should display the model's statistics values, not NaN.
Version sasctl==1.10.6