Model settings

Under model settings, you can find all the monitoring parameters of a selected model. These settings are specific to a single model and are not confused with the general NannyML settings in the navbar.

On the left side, you can navigate through the different configuration groupings. There is also a "Run now" button to trigger a new NannyML run. This might be useful after some of the parameters are updated.

General details

Here, you can change the name of your model.

Datasets

Under datasets, you can manually add more analysis and target data.

Schedule

Under schedule, you can define when to run the drift and metric calculators.

Chunking

Here, you can choose how to group the results by time interval or size. For example, choosing "monthly" groups all predictions made in the same month and calculates the results.

💡 We currently only support time-based and size-based chunking; if you need support for number-based chunking, contact us.

Performance

Here, you can select the metrics you want to monitor. There is also the option to configure them further. The metrics will either be calculated and/or estimated depending on the selected performance types. Calculating metrics and thus measuring realized performance is only possible if targets are supplied.

Under every metric configuration, it is possible to specify further if this metric has to be calculated and/or estimated. NannyML automatically extracts thresholds based on the supplied reference data, but it is possible to configure a custom threshold here. All metrics follow this type of configuration except business value.

There are two types of threshold constants and standard deviation-based thresholds:

For business value estimation or calculation, a cost/benefit matrix has to be supplied. This matrix contains the value a single observation in each of the cells of the respective confusion matrix cells brings in or costs. For example, a true positive prediction brings in X amount, and a false positive prediction will cost us Y.

Concept shift

Here, you can specify which concept shift results to run and configure the threshold values.

Covariate shift

In covariate shift settings, you can specify which drift methods to run and also configure the threshold values.

Some methods work for categorical and continuous columns; in that case, it can be selected which of those they must run. Also, the threshold can be manually configured.

Data quality

Here, you can select the type of data quality checks and their threshold values.

Both missing values and unseen values can be normalized along with default thresholds.