Quick start: Evaluate a machine learning model

Here are some tips for successfully completing this tutorial.

This tutorial uses a sample project containing a machine learning model and a notebook to configure the monitors. Follow these steps to create a project based on a sample.

1. Download the evaluate-an-ml-model.zip file.

2. From the navigation menu , choose Projects > All projects.

3. On the Projects page, click New project.

4. Select Create a project from a file.

5. Upload the previously downloaded ZIP file.

6. On the Create a project page, copy and paste the project name and add an optional description for the project.

   Evaluate an ML model
   

7. Click Create.

8. Click View new project to verify that the project and assets were created successfully.

9. Click the Assets tab to view the project's assets.

Before you can deploy the model, you need to promote the model to a new deployment space. Deployment spaces help you to organize supporting resources such as input data and environments; deploy models or functions to generate predictions or solutions; and view or edit deployment details.

Run the second notebook included in the sample project to:

• Fetch the model and deployments.
• Configure Watson OpenScale.
• Create the service provider and subscription for your machine learning service.
• Configure the quality monitor.
• Configure the fairness monitor.
• Configure explainability.

Follow these steps to run the notebook included in the sample project. This notebook sets up monitors for your model, which can also be configured through the user interface. However, it is quicker and less error prone to set them up with a notebook. Take some time to read through the comments in the notebook, which explain the code in each cell.

1. From the navigation menu , choose Projects > All projects.

2. Open the Evaluate an ML model project.

3. Click the Assets tab, and then navigate to Notebooks.
   

4. Open the monitor-wml-model-with-watson-openscale notebook.

5. Since the notebook is in read-only mode, click the Edit icon to place the notebook in edit mode.

6. The WOS Credentials cell requires your input.

   1. At the Enter host name prompt, type your Cloud Pak for Data hostname beginning with https://, and press Enter. For example, https://mycpdcluster.mycompany.com.

   2. At the Username prompt, type your Cloud Pak for Data username, and press Enter.

   3. At the Password prompt, type your Cloud Pak for Data password, and press Enter.

7. In the 3. Model and Deployment section, verify the values assigned to the space_name, model_name, and deployment_name variables.

8. Click Cell > Run All to run all of the cells in the notebook. Alternatively, click the Run icon next to each cell to run the notebook cell by cell to explore each cell and its output.

9. The notebook takes 1 - 3 minutes to complete. You can monitor the progress cell by cell by noticing the asterisk "In [*]" changing to a number, for example, "In [1]".

10. If you encounter any errors during the notebook run, try these troubleshooting tips:

    • Click Kernel > Restart & Clear Output to restart the kernel, and then run the notebook again.

    • Delete any existing Watson OpenScale deployments, and provision a new service instance.

    • Verify that you created the deployment space and deployment name by copying and pasting the specified artifact name exactly with no leading or trailing spaces.

Follow these steps to download holdout data, and use that data to evaluate the model in Watson OpenScale:

1. Click the Evaluate an ML model project in the navigation trail.

   

2. On the Assets tab, click Data > Data assets.

3. Click the Overflow menu for the GoldenBank_HoldoutData.csv data asset, and choose Download. To validate that the model is working as required, you need a set of labeled data, which was held out from model training. This CSV file contains that holdout data.

4. Launch Watson OpenScale.

   1. From the navigation menu , choose Services > Service instances.

   2. On the Instances page, click the Overflow menu at the end of the row for your Watson OpenScale instance, and choose Open.

5. On the Insights dashboard, click the Mortgage Approval Model Deployment tile.

6. From the Actions menu, select Evaluate now.

7. From the list of import options, select from CSV file.

8. Drag the Golden Bank_HoldoutData.csv data file you downloaded from the project into the side panel.

9. Click Upload and evaluate and wait for the evaluation to complete.

The Watson OpenScale quality monitor generates a set of metrics to evaluate the quality of your model. You can use these quality metrics to determine how well your model predicts outcomes. When the evaluation that uses the holdout data completes, follow these steps to observe the model quality or accuracy:

1. In the Quality section, click the Configure icon. Here you can see that the quality threshold that is configured for this monitor is 70% and that the measurement of quality being used is area under the ROC curve.

2. Click Go to model summary to return to the model details screen.

3. In the Quality section, click the right arrow icon to see the model quality detailed results. Here you see a number of quality metric calculations and a confusion matrix showing correct model decisions along with false positives and false negatives. The calculated area under the ROC curve is 0.9 or higher, which exceeds the 0.7 threshold, so the model is meeting its quality requirement.

4. Click Mortgage Approval Model Deployment in the navigation trail to return to the model details screen.

The Watson OpenScale fairness monitor generates a set of metrics to evaluate the fairness of your model. You can use the fairness metrics to determine if your model produces biased outcomes. Follow these steps to observe the model fairness:

1. In the Fairness section, click the Configure icon. Here you see that the model is being reviewed to ensure that applicants are being treated fairly regardless of their gender. Women are identified as the monitored group for whom fairness is being measured and the threshold for fairness is to be at least 80%. The fairness monitor uses the disparate impact method to determine fairness. Disparate impact compares the percentage of favorable outcomes for a monitored group to the percentage of favorable outcomes for a reference group.

2. Click Go to model summary to return to the model details screen.

3. In the Fairness section, click the right arrow icon to see the model fairness detailed results. Here you see the percentage of male and female applicants who are being automatically approved, along with a fairness score of over 100%, so the model performance far exceeds the 80% fairness threshold required.

4. Note the identified data sets. To ensure that the fairness metrics are most accurate, Watson OpenScale uses perturbation to determine the results where only the protected attributes and related model inputs are changed while other features remain the same. The perturbation changes the values of the feature from the reference group to the monitored group, or vice-versa. These additional guardrails are used to calculate fairness when the "balanced" data set is used, but you can also view the fairness results using only payload or model training data. Since the model is behaving fairly, you don't need to go into additional detail for this metric.

   

5. Click the Mortgage Approval Model Deployment navigation trail to return to the model details screen.

The Watson OpenScale drift monitor measures changes in your data over time to ensure consistent outcomes for your model. Use drift evaluations to identify changes in your model output, the accuracy of your predictions, and the distribution of your input data. Follow these steps to observe the model drift:

1. In the Drift section, click the Configure icon. Here you see the drift thresholds. Output drift measures the change in the model confidence distribution. Model quality drift measures the drop in accuracy by comparing the estimated runtime accuracy to the training accuracy. Feature drift measures the change in value distribution for important features. The configuration also shows the number of selected features and the most important features.

2. Click Go to model summary to return to the model details screen.

3. In the Drift section, click the right arrow icon to see the model drift detailed results. You can view the history of how each metric score changes over time with a time series chart. Lower values are better, so in this case, the results are above the upper thresholds that are set in the configuration. Then view details about how the scores output and feature drifts are calculated. You can also view details about each feature to understand how they contribute to the scores that Watson OpenScale generates.

4. Click the Mortgage Approval Model Deployment navigation trail to return to the model details screen.

It is also important to understand how the model came to its decision. This understanding is required both to explain decisions to people involved in the loan approval and to ensure model owners that the decisions are valid. To understand these decisions, follow these steps to observe the model explainability:

1. In the left navigation panel, click the Explain a transaction icon.

2. Select Mortgage Approval Model Deployment to see a list of transactions.

3. For any transaction, click Explain under the Actions column. Here you see the detailed explanation of this decision. You will see the most important inputs to the model along with how important each was to the end result. Blue bars represent inputs that tended to support the model's decision while red bars show inputs that might have led to another decision. For example, an applicant might have enough income to otherwise be approved but their poor credit history and high debt together lead the model to reject the application. Review this explanation to become satisfied about the basis for the model decision.

4. (Optional) If you want to delve further into how the model made its decision, click the Inspect tab. Use the Inspect feature to analyze the decision to find areas of sensitivity where a small changes to a few inputs would result in a different decision, and you can test the sensitivity yourself by overriding some of the actual inputs with alternatives to see whether these would impact the result.