Analytics Manager > Working with Predictive Scoring > How Predictive Scoring Works Using the Analytics Server Connector > Troubleshoot Predictive Scoring Using the Analytics Server
  
Troubleshoot Predictive Scoring Using the Analytics Server
The following table provides information that helps you troubleshoot scenarios that you might encounter while performing predictive scoring using the Analytics Server Connector:
Scenario
Error Message
Solution
If you enter invalid input values while creating an analysis job, the job goes into the COMPLETED state. An error message is returned from the prediction job and it is written back to the errorMessage property.
For example: If you enter invalid values for the Age Group and the ZipCode fields, then you get the following error message:
The row failed to evaluate due to Unexpected value(s) found in the record. Field [Age Group] had value of []; Field [ZipCode] had value of []
Enter valid input values.
Consider the following scenario:
1. You have configured an Analytics Server in Analytics Builder.
2. You have uploaded a model to the configured Analytics Server.
3. Now, you configure a different Analytics Server in Analytics Builder.
4. If you try to create an analysis job for the model, the job is created, but it goes in the FAILED state.
This model was uploaded using the <OLD_ANALYTICS_SERVER>Analytics Server. However, the current configured Analytics Server is <CURRENT_ANALYTICS_SERVER>. You can disable the event that is configured on this model, and create a model on the current Analytics Server.
Solution 1
1. Upload the model on the current configured Analytics Server
2. Create an analysis job.
Solution 2
1. Configure the same Analytics Server in Analytics Builder that you had uploaded the model to.
2. Create an analysis job.
For time series models, if the number of input data rows is less than the lookback count, then the scoring job fails. An error message is returned from the prediction job and it is written to the errorMessage property.
Cannot perform a time series transformation when the data count [<DATA_COUNT>] is less than the sum of the lookback window size [<LOOKBACK_WINDOW_SIZE>] and the lookahead [<LOOKAHEAD_COUNT>], which is equal to [<LOOKBACK_WINDOW_SIZE>+<LOOKAHEAD_COUNT>]
Provide appropriate number of input data rows with appropriate input values so that a scoring job can be executed.
For time series models, if you specify the value of the important field count field to be greater than 0, then the scoring job fails.
Time series does not support an important field count greater than 0. A value of [<SPECIFIED_IMPORTANT_FIELD_COUNT>] was specified in the request.
Specify the value of the important field count as 0.