piwb.underperformance_min_records The piwb.underperformance_min_records property sets the minimum number of predictions required to assess underperformance in use cases. Admins should review this setting to ensure accurate evaluations, especially when dealing with varying data volumes.
1000The answers below summarize the purpose, scope, default effect, and review scenarios for this property.
| What does it do? | This property defines the threshold for the minimum predictions needed to evaluate the underperformance of use cases, impacting data analysis accuracy. |
|---|---|
| What area does it affect? | Predictive Analytics |
| What does the default mean? | The default value is "1000", which sets the baseline numeric behavior for this property. |
| When should you review it? | Review this property when adjusting prediction thresholds or analyzing use case performance. |
Raw metadata from the property record.
| Property name | piwb.underperformance_min_records |
|---|---|
| Sys ID | f250c7f267141010705c82f45685ef32 |
| Type | integer |
| Application | Global |
| Default value | 1000 |
| Description | The minimum # of predictions for evaluating use case underperformance |
| Updated | 2020-04-24 20:34:29 |
Official references and nearby text excerpts where this property appears.
piwbOther properties in the same namespace.
piwb.instance_eligible_auto_train
This property indicates the eligibility of an instance for pretraining, affecting its training capabilities.
piwb.prediction_batch_size
This property defines the limit on the number of rows processed in each batch for PIWB Model predictions, impa
piwb.prediction_max_allowed_rows
This property limits the number of rows processed in batch testing to enhance efficiency and manage resource u
piwb.top_n_input_fields
This property configures the count of recommended input fields displayed based on their correlation to the out
piwb.underperformance_threshold
This property defines the acceptable percentage difference that indicates underperformance in automation use c
piwb.auto_train_threshold
This property defines the threshold percentage of net automation necessary for a pretrained model to be visibl