Model Drift Detection with Assays

Wallaroo provides the ability to monitor the input and outputs of models to detect model drift by using model insights. Changes in the inputs or model predictions that fall outside of expected norms, known as data drift, can occur due to errors in the data processing pipeline or due to changes in the environment such as user preference or behavior.

Monitoring tasks called assays provides model insights by comparing a model’s predictions or the input data coming into the model against an established baseline. Changes in the distribution of this data can be an indication of model drift, or of a change in the environment that the model trained for. This can provide tips on whether a model needs to be retrained or the environment data analyzed for accuracy or other needs.

Model Insights provides interactive assays to explore data from a pipeline and learn how the data is behaving. With this information and the knowledge of your particular business use case you can then choose appropriate settings to launch an assay to run on a given frequency as more data is collected.

Assay versions

Wallaroo defaults to Assay V1 and introduced Assays V2 in Wallaroo 2024.2. Assays V2 provides additional performance enhancements and multi-threaded processing of assay runs for realtime drift detection scenarios and offline drift detection scenarios. Except where noted, the Wallaroo SDK and Wallaroo MLOps API calls and returns are the same.

Configuring Assay Version

The assay version is set through platform operations, either during installation or via update.

When changing from one assay version to the other, the following effects will follow:

• The Wallaroo Dashboard Pipeline Insights page only displays assays created in the configured assay version.
  • For example, if the assay was created in Assays V1, then the Wallaroo instance is configured to Assays V2, only assays created under Assays V2 are displayed in the Wallaroo Dashboard.
• The Wallaroo SDK Assay methods must be set to same configured Wallaroo Assay version.
• The Wallaroo MLOps API endpoints only support Assays V1.

Organizations should inform ML engineers and other developers on what Wallaroo Assay version is used so they can use the correct API and SDK settings.

It is highly recommended to set the Wallaroo Assay version at installation time, or before any assays are created. When moving from one Wallaroo Assay version to another, all existing assays should be migrated by recreating them with the same original settings.

Model Drift Detection with Model Insights via the Wallaroo Dashboard

Model Drift Detection with Model Insights via the Wallaroo Dashboard allows organizations to create assays that monitor inference requests and results against an established baseline.

Access the Model Insights Page

Assays created through the Wallaroo Dashboard are accessed through the Pipeline Dashboard through the following process.

1. Log into the Wallaroo Dashboard.
2. Select the workspace containing the pipeline with the models being monitored from the Change Current Workspace and Workspace Management drop down.
3. Select View Pipelines.
4. Select the pipeline containing the models being monitored.
5. Select Insights.

The Wallaroo Assay Dashboard contains the following elements. For more details of each configuration type, see the Model Insights and Assays Introduction.

• Pipeline Name and Identifier (A): The pipeline’s assigned name and unique identifier in UUID format.

• Filter Edges (B): By default, all locations are displayed. Filter Edges provides a list of available edge deployments are displayed. Selecting one or more filters from the list limits the available metrics and logs displayed to only the selected locations.

  

• Status (C): The status of the pipeline. The status only applies to the pipeline’s status in the Wallaroo Ops instance.. Options are:

  • Active: The pipeline is deployed.
  • Inactive: The pipeline is not deployed.

• Tags (D): Any tags applied to the pipeline.

• Inference Urls (E): The internal and external inference URLs for the deployed pipeline in the Wallaroo Ops instance.

• + Create Assay (F): Create a new assay.

• Deploy/Undeploy the Pipeline (G): Deploy an inactive pipeline, or deploy an active pipeline in the Wallaroo Ops instance.

• Filter Assays (H): Filter the assays by Name or Status, and Sort.

  • Status filters are:
    • All Status: All statuses.
    • Active: Active assays.
    • Paused: Assays that are paused.
    • Drift Detected: Assays where model drift is detected.
  • Sort filters are:
    • Creation Date
    • Last Assay Run

• Assay Details View (I): A chart of the assay analyses. This is filtered by the selected dates and time ranges.

• Activity (J): Comments left by users.

Assay Details View

The following details are visible by selecting the Assay View Details icon:

• (A) Assay Name: The name of the assay displayed.
• (B) Input / Output: The input or output and the index of the element being monitored.
• (C) Baseline: The time period used to generate the baseline. For baselines generated from a file, the baseline displayed Uploaded File.
• (D) Last Run: The date and time the assay was last run.
• (E) Next Run: The future date and time the assay will be run again. NOTE: If the assay is paused, then it will not run at the scheduled time. When unpaused, the date will be updated to the next date and time that the assay will be run.
• (F) Aggregation Type: The aggregation type used with the assay.
• (G) Threshold: The threshold value used for the assay.
• (H) Metric: The metric type used for the assay.
• (I) Number of Bins: The number of bins used for the assay.
• (J) Bin Weight: The weight applied to each bin.
• (K) Bin Mode: The type of bin node applied to each bin.

View Assay Alert Details

To view details on an assay alert:

1. Select the data with available alert data.
2. Mouse hover of a specific Assay Event Alert to view the data and time of the event and the alert value.
3. Select the Assay Event Alert to view the Baseline and Window details of the alert including the left_outlier and right_outlier.

Hover over a bar chart graph to view additional details.

1. Select the ⊗ symbol to exit the Assay Event Alert details and return to the Assay View.

Build an Assay Through the Pipeline Dashboard

To create a new assay through the Wallaroo Pipeline Dashboard:

1. Log into the Wallaroo Dashboard.
2. Select the workspace containing the pipeline with the models being monitored from the Change Current Workspace and Workspace Management drop down.
3. Select View Pipelines.
4. Select the pipeline containing the models being monitored.
5. Select Insights.
6. Select +Create Assay.
7. On the Create Assay module, enter the following:

   1. On the Assay Name section, enter the following:

      

   2. Assay Name (A): The name of the new assay.

   3. Monitor output data or Monitor input data (B): Select whether to monitor input or output data.

   4. Select an output/input to monitor (C): Select the input or output to monitor.

      1. Named Field: The name of the field to monitor.
      2. Index: The index of the monitored field.

   5. On the Specify Baseline section, select one of the following options:

      1. (D) Select the data to use for the baseline. This can either be set with a preset recent time period (last 30 seconds, last 60 seconds, etc) or with a custom date range.

      

      1. (E) Upload an assay baseline file as either a CSV or TXT file. These assay baselines must be a list of numpy (aka float) values that are comma and newline separated, terminating at the last record with no additional commas or returns.

         

         For example:

         684577.200,
         921561.500,
         705013.440,
         725875.900,
         684577.200,
         379398.300,
         266405.600,
         256630.310
         

      Once selected, a preview graph of the baseline values will be displayed (C). Note that this may take a few seconds to generate.

   6. Select Next to continue.

8. On the Settings Module:

   1. Set the date and time range to view values generated by the assay. This can either be set with a preset recent time period (last 30 seconds, last 60 seconds, etc) or with a custom date range.

      New assays are configured to run a new analysis for every 24 hours starting at the end of the baseline period. For information on how to adjust the scheduling period and other settings for the assay scheduling window, see the SDK section on how to Schedule Assay.

   2. Set the following Advanced Settings.

      

      1. (A) Preview Date Range: The date and times to for the preview chart.
      2. (B) Preview: A preview of the assay results will be displayed based on the settings below.
      3. (C) Scheduling: Set the Frequency (Daily, Every Minute, Hourly, Weekly, Default: Daily) and the Time (increments of one hour Default: 1:00 AM).
      4. (D) Group Results: How the results are grouped: Daily (Default), Every Minute, Weekly, or Monthly.
      5. (E) Aggregation Type: Density or Cumulative.
      6. (F) Threshold:
         1. Default: 0.1
      7. (G) Metric:
         1. Default: Population Stability Index
         2. Maximum Difference of Bins
         3. Sum of the Difference of Bins
      8. (H) Number of Bins: From 5 to 14. Default: 5
      9. (F) Bin Mode:
         1. Equally Spaced
         2. Default: Quantile
      10. (I) Bin Weights: The bin weights:
          1. Equally Weighted (Default)
          2. Custom: Users can assign their own bin weights as required.

   3. Review the preview chart to verify the settings are correct.

   4. Select Build to complete the process and build the new assay.

Once created, it may take a few minutes for the assay to complete compiling data. If needed, reload the Pipeline Dashboard to view changes.

Model Insights via the Wallaroo Dashboard SDK

Assays generated through the Wallaroo SDK can be previewed, configured, and uploaded to the Wallaroo Ops instance. The following is a condensed version of this process. For full details see the Wallaroo SDK Essentials Guide: Assays Management guide.

Model drift detection with assays using the Wallaroo SDK follows this general process.

• Define the Baseline: From either historical inference data for a specific model in a pipeline, or from a pre-determine array of data, a baseline is formed.
• Assay Preview: Once the baseline is formed, we preview the assay and configure the different options until we have the the best method of detecting environment or model drift.
• Create Assay: With the previews and configuration complete, we upload the assay. The assay will perform an analysis on a regular scheduled based on the configuration.
• Get Assay Results: Retrieve the analyses and use them to detect model drift and possible sources.
• Pause/Resume Assay: Pause or restart an assay as needed.

Define the Baseline

Assay baselines are defined with the wallaroo.client.build_assay method. Through this process we define the baseline from either a range of dates or pre-generated values.

wallaroo.client.build_assay take the following parameters:

Note that model_name is an optional parameters when parameters are named. For example:

assay_builder_from_dates = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

or:

assay_builder_from_dates = wl.build_assay("assays from date baseline", 
                                          mainpipeline, 
                                          None, ## since we are using positional parameters, `None` must be included for the model parameter
                                          "output variable 0",
                                          assay_baseline_start, 
                                          assay_baseline_end)

Baselines are created in one of two mutually exclusive methods:

• Date Range: The baseline_start and baseline_end retrieves the inference requests and results for the pipeline from the start and end period. This data is summarized and used to create the baseline. For our examples, we’re using the variables assay_baseline_start and assay_baseline_end to represent a range of dates, with assay_baseline_start being set before assay_baseline_end.
• Numpy Values: The baseline_data sets the baseline from a provided numpy array. This allows assay baselines to be created without first performing inferences in Wallaroo.

Define the Baseline Example

This example shows two methods of defining the baseline for an assay:

• "assays from date baseline": This assay uses historical inference requests to define the baseline. This assay is saved to the variable assay_builder_from_dates.
• "assays from numpy": This assay uses a pre-generated numpy array to define the baseline. This assay is saved to the variable assay_builder_from_numpy.

In both cases, the following parameters are used:

The difference between the two assays’ parameters determines how the baseline is generated.

• "assays from date baseline": Uses the baseline_start and baseline_end to set the time period of inference requests and results to gather data from.
• "assays from numpy": Uses a pre-generated numpy array as for the baseline data.

First we generate an assay baseline from a range of historical inferences performed through the specified pipeline deployment.

# Build the assay, based on the start and end of our baseline time, 
# and tracking the output variable index 0

display(assay_baseline_start)
display(assay_baseline_end)

assay_baseline_from_dates = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# create the baseline from the dates
assay_baseline_run_from_dates = assay_baseline_from_dates.build().interactive_baseline_run()

datetime.datetime(2024, 10, 18, 15, 50, 21, 296181)

datetime.datetime(2024, 10, 18, 15, 51, 21, 378013)

# storing the workspace id for later requests
workspace_id = workspace.id()
workspace_id

11

In this code sample, we generate the assay baseline from a preset numpy array called small_results_baseline.

# build the baseline from the numpy array
display(small_results_baseline[0:5])

# assay builder by baseline
assay_baseline_from_numpy = wl.build_assay(assay_name="assays from numpy", 
                               pipeline=mainpipeline, 
                               iopath="output variable 0", 
                               baseline_data = small_results_baseline)

# create the baseline from the numpy array
assay_baseline_run_from_numpy = assay_baseline_from_numpy.build().interactive_baseline_run()

array([ 450867.7 , 1140733.8 ,  974485.3 ,  873848.44,  249227.83])

Baseline Chart

The baseline chart is displayed with wallaroo.assay.AssayAnalysis.chart(), which returns a chart with:

• baseline mean: The mean value of the baseline values.
• baseline median: The median value of the baseline values.
• bin_mode: The binning mode. See Binning Mode
• aggregation: The aggregation type. See Aggregation Options
• metric: The assay’s metric type. See Score Metric
• weighted: Whether the binning mode is weighted. See Binning Mode

The first chart is from an assay baseline generated from a set of inferences across a range of dates.

assay_baseline_run_from_dates.chart()

baseline mean = 539895.63075
baseline median = 450283.78125
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False

This chart is from an assay baseline generated from a set of numpy values.

assay_baseline_run_from_numpy.chart()

baseline mean = 539895.6332000002
baseline median = 450283.79
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False

Baseline DataFrame

The method wallaroo.assay_config.AssayBuilder.baseline_dataframe returns a DataFrame of the assay baseline generated from the provided parameters. This includes:

• metadata: The inference metadata with the model information, inference time, and other related factors.
• in data: Each input field assigned with the label in.{input field name}.
• out data: Each output field assigned with the label out.{output field name}

Note that for assays generated from numpy values, there is only the out data based on the supplied baseline data.

In the following example, the baseline DataFrame is retrieved.

This baseline DataFrame is from an assay baseline generated from a set of inferences across a range of dates.

display(assay_baseline_from_dates.baseline_dataframe())

500 rows × 21 columns

This baseline DataFrame is from an assay baseline generated from an array of numpy values.

display(assay_baseline_from_numpy.baseline_dataframe())

500 rows × 1 columns

Baseline Stats

The method wallaroo.assay.AssayAnalysis.baseline_stats() returns a pandas.core.frame.DataFrame of the baseline stats.

The baseline stats for each assay are displayed in the examples below.

This baseline states DataFrame is from an assay baseline generated from a set of inferences across a range of dates.

assay_baseline_run_from_dates.baseline_stats()

This baseline states DataFrame is from an assay baseline generated from an array of numpy values.

assay_baseline_run_from_numpy.baseline_stats()

Baseline Bins

The method wallaroo.assay.AssayAnalysis.baseline_bins a simple dataframe to with the edge/bin data for a baseline.

These baseline bins DataFrame is from an assay baseline generated from a set of inferences across a range of dates.

assay_baseline_run_from_dates.baseline_bins()

These baseline bins DataFrame is from an assay baseline generated from an array of numpy values.

assay_baseline_run_from_numpy.baseline_bins()

Baseline Histogram Chart

The method wallaroo.assay_config.AssayBuilder.baseline_histogram returns a histogram chart of the assay baseline generated from the provided parameters.

These chart is from an assay baseline generated from a set of inferences across a range of dates.

assay_baseline_from_dates.baseline_histogram()

Baseline KDE Chart

The method wallaroo.assay_config.AssayBuilder.baseline_kde returns a Kernel Density Estimation (KDE) chart of the assay baseline generated from the provided parameters.

These chart is from an assay baseline generated from a set of inferences across a range of dates.

assay_baseline_from_dates.baseline_kde()

Baseline ECDF Chart

The method wallaroo.assay_config.AssayBuilder.baseline_ecdf returns a Empirical Cumulative Distribution Function (CDF) chart of the assay baseline generated from the provided parameters.

These chart is from an assay baseline generated from a set of inferences across a range of dates.

assay_baseline_from_dates.baseline_ecdf()

Assay Preview

Now that the baseline is defined, we look at different configuration options and view how the assay baseline and results changes. Once we determine what gives us the best method of determining model drift, we can create the assay.

Analysis List Chart Scores

Analysis List scores show the assay scores for each assay result interval in one chart. Values that are outside of the alert threshold are colored red, while scores within the alert threshold are green.

Assay chart scores are displayed with the method wallaroo.assay.AssayAnalysisList.chart_scores, with ability to display an optional title with the chart. This takes the following parameters:

The following example shows retrieving the assay results and displaying the chart scores. From our example, we have two windows - the first should be green, and the second is red showing that values were outside the alert threshold.

assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show the analyses chart
assay_results.chart_scores()

The following chart uses the nth_x_tick setting to space out the x-axis labels further.

# set different nth_x_tick
assay_results.chart_scores(nth_x_tick=10)

Analysis Chart

The method wallaroo.assay.AssayAnalysis.chart() displays a comparison between the baseline and an interval of inference data.

This is compared to the Chart Scores, which is a list of all of the inference data split into intervals, while the Analysis Chart shows the breakdown of one set of inference data against the baseline.

Score from the Analysis List Chart Scores and each element from the Analysis List DataFrame generates

The following fields are included.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline sample", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display one of the analysis from the total results
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False
score = 0.027998899384779063
scores = [0.0, 0.00031071866653010913, 0.00030773024662236867, 0.002769365409558061, 0.015958199760662362, 0.008652885301406161, 0.0]
index = None

Analysis List DataFrame

wallaroo.assay.AssayAnalysisList.to_dataframe() returns a DataFrame showing the assay results for each window aka individual analysis. This DataFrame contains the following fields:

For this example, the assay analysis list DataFrame is listed.

From this tutorial, we should have 2 windows of dta to look at, each one minute apart. The first window should show status: OK, with the second window with the very large house prices will show status: alert

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display the dataframe from the analyses
assay_results.to_dataframe()

Analysis List Full DataFrame

wallaroo.assay.AssayAnalysisList.to_full_dataframe() returns a DataFrame showing all values, including the inputs and outputs from the assay results for each window aka individual analysis. This DataFrame contains the following fields:

pipeline_id	warning_threshold	bin_index	created_at

For this example, full DataFrame from an assay preview is generated.

From this tutorial, we should have 2 windows of dta to look at, each one minute apart. The first window should show status: OK, with the second window with the very large house prices will show status: alert

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display the full dataframe from the analyses
assay_results.to_full_dataframe()

Analysis Compare Basic Stats

The method wallaroo.assay.AssayAnalysis.compare_basic_stats returns a DataFrame comparing one set of inference data against the baseline.

This is compared to the Analysis List DataFrame, which is a list of all of the inference data split into intervals, while the Analysis Compare Basic Stats shows the breakdown of one set of inference data against the baseline.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display one analysis against the baseline
assay_results[0].compare_basic_stats()

Configure Assays

Before creating the assay, configure the assay and continue to preview it until the best method for detecting drift is set. The following options are available.

Inference Interval and Inference Width

The inference interval aka window interval sets how often to run the assay analysis. This is set from the wallaroo.assay_config.AssayBuilder.window_builder.add_interval method to collect data expressed in time units: “hours=24”, “minutes=1”, etc.

For example, with an interval of 1 minute, the assay collects data every minute. Within an hour, 60 intervals of data is collected.

We can adjust the interval and see how the assays change based on how frequently they are run.

The width sets the time period from the wallaroo.assay_config.AssayBuilder.window_builder.add_width method to collect data expressed in time units: “hours=24”, “minutes=1”, etc.

For example, an interval of 1 minute and a width of 1 minute collects 1 minutes worth of data every minute. An interval of 1 minute with a width of 5 minutes collects 5 minute of inference data every minute.

By default, the interval and width is 24 hours.

For this example, we’ll adjust the width and interval from 1 minute to 5 minutes and see how the number of analyses and their score changes.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show the analyses chart
assay_results.chart_scores()

wl.get_current_workspace()

{'name': 'assay-demonstration-tutorial-2', 'id': 11, 'archived': False, 'created_by': '3eeab554-fc2e-41ac-8b4e-0d778e8a471b', 'created_at': '2024-10-09T19:50:23.231562+00:00', 'models': [{'name': 'house-price-estimator', 'versions': 3, 'owner_id': '""', 'last_update_time': datetime.datetime(2024, 10, 18, 15, 50, 20, 532959, tzinfo=tzutc()), 'created_at': datetime.datetime(2024, 10, 9, 19, 50, 24, 679975, tzinfo=tzutc())}], 'pipelines': [{'name': 'assay-demonstration-tutorial', 'create_time': datetime.datetime(2024, 10, 9, 19, 50, 25, 224073, tzinfo=tzutc()), 'definition': '[]'}]}

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to five minutes each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show the analyses chart
assay_results.chart_scores()

Add Run Until and Add Inference Start

For previewing assays, setting wallaroo.assay_config.AssayBuilder.add_run_until sets the end date and time for collecting inference data. When an assay is uploaded, this setting is no longer valid - assays run at the Inference Interval until the assay is paused.

Setting the wallaroo.assay_config.WindowBuilder.add_start sets the start date and time to collect inference data. When an assay is uploaded, this setting is included, and assay results will be displayed starting from that start date at the Inference Interval until the assay is paused. By default, add_start begins 24 hours after the assay is uploaded unless set in the assay configuration manually.

For the following example, the add_run_until setting is set to datetime.datetime.now() to collect all inference data from assay_window_start up until now, and the second example limits that example to only two minutes of data.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results, minus 2 minutes for the period to start gathering analyses
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start+datetime.timedelta(seconds=-120))

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show the analyses chart
assay_results.chart_scores()

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results plus 2 minutes 
assay_baseline_from_dates.add_run_until(assay_window_start+datetime.timedelta(seconds=120))

# Set the interval and window to one minute each
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show the analyses chart
assay_results.chart_scores()

Score Metric

The score is a distance between the baseline and the analysis window. The larger the score, the greater the difference between the baseline and the analysis window. The following methods are provided determining the score:

• PSI (Default) - Population Stability Index (PSI).
• MAXDIFF: Maximum difference between corresponding bins.
• SUMDIFF: Mum of differences between corresponding bins.

The metric type used is updated with the wallaroo.assay_config.AssayBuilder.add_metric(metric: wallaroo.assay_config.Metric) method.

The following three charts use each of the metrics. Note how the scores change based on the score type used.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# set metric PSI mode
assay_baseline.summarizer_builder.add_metric(wallaroo.assay_config.Metric.PSI)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display one analysis from the results
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False
score = 0.027998899384779063
scores = [0.0, 0.00031071866653010913, 0.00030773024662236867, 0.002769365409558061, 0.015958199760662362, 0.008652885301406161, 0.0]
index = None

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# set metric MAXDIFF mode
assay_baseline.summarizer_builder.add_metric(wallaroo.assay_config.Metric.MAXDIFF)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display one analysis from the results
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = MaxDiff
weighted = False
score = 0.061000000000000026
scores = [0.0, 0.00799999999999998, 0.008000000000000007, 0.021999999999999992, 0.061000000000000026, 0.03900000000000001, 0.0]
index = 4

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# set metric SUMDIFF mode
assay_baseline.summarizer_builder.add_metric(wallaroo.assay_config.Metric.SUMDIFF)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# display one analysis from the results
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = SumDiff
weighted = False
score = 0.069
scores = [0.0, 0.00799999999999998, 0.008000000000000007, 0.021999999999999992, 0.061000000000000026, 0.03900000000000001, 0.0]
index = None

Alert Threshold

Assay alert thresholds are modified with the wallaroo.assay_config.AssayBuilder.add_alert_threshold(alert_threshold: float) method. By default alert thresholds are 0.1.

The following example updates the alert threshold to 0.5.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# set the alert threshold
assay_baseline.add_alert_threshold(0.5)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show the analyses with the alert threshold
assay_results.to_dataframe()

Number of Bins

Number of bins sets how the baseline data is partitioned. The total number of bins includes the set number plus the left_outlier and the right_outlier, so the total number of bins will be the total set + 2.

The number of bins is set with the wallaroo.assay_config.UnivariateContinousSummarizerBuilder.add_num_bins(num_bins: int) method.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# update number of bins here
assay_baseline.summarizer_builder.add_num_bins(10)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show one analysis with the updated bins
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False
score = 0.05908866707776886
scores = [0.0, 0.005162673110806691, 0.0027300082724922136, 0.001619315394342909, 0.0042755780600660315, 0.0015054510098891697, 0.0012818719788154692, 0.006791725840796723, 0.009249964850087234, 0.0002488075477953185, 0.026223271012677105, 0.0]
index = None

Binning Mode

Binning Mode defines how the bins are separated. Binning modes are modified through the wallaroo.assay_config.UnivariateContinousSummarizerBuilder.add_bin_mode(bin_mode: bin_mode: wallaroo.assay_config.BinMode, edges: Optional[List[float]] = None).

Available bin_mode values from wallaroo.assay_config.Binmode are the following:

• QUANTILE (Default): Based on percentages. If num_bins is 5 then quintiles so bins are created at the 20%, 40%, 60%, 80% and 100% points.
• EQUAL: Evenly spaced bins where each bin is set with the formula min - max / num_bins
• PROVIDED: The user provides the edge points for the bins.

If PROVIDED is supplied, then a List of float values must be provided for the edges parameter that matches the number of bins.

The following examples are used to show how each of the binning modes effects the bins.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# update binning mode here
assay_baseline.summarizer_builder.add_bin_mode(wallaroo.assay_config.BinMode.QUANTILE)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show one analysis with the updated bins
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False
score = 0.027998899384779063
scores = [0.0, 0.00031071866653010913, 0.00030773024662236867, 0.002769365409558061, 0.015958199760662362, 0.008652885301406161, 0.0]
index = None

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# update binning mode here
assay_baseline.summarizer_builder.add_bin_mode(wallaroo.assay_config.BinMode.EQUAL)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show one analysis with the updated bins
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Equal
aggregation = Density
metric = PSI
weighted = False
score = 0.024337608548999315
scores = [0.0, 0.0004539311540674428, 0.0102623545559447, 0.0015340004581186216, 0.00863998698518399, 0.003447335395684558, 0.0]
index = None

The following example manually sets the bin values.

The values in this dataset run from 200000 to 1500000. We can specify the bins with the BinMode.PROVIDED and specifying a list of floats with the right hand / upper edge of each bin and optionally the lower edge of the smallest bin. If the lowest edge is not specified the threshold for left outliers is taken from the smallest value in the baseline dataset.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

edges = [200000.0, 400000.0, 600000.0, 800000.0, 1500000.0, 2000000.0]

# update binning mode here
assay_baseline.summarizer_builder.add_bin_mode(wallaroo.assay_config.BinMode.PROVIDED, edges)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show one analysis with the updated bins
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Provided
aggregation = Density
metric = PSI
weighted = False
score = 0.022475023501744996
scores = [0.0, 0.0012824905633907766, 0.0004230661205221289, 0.007031010052631133, 0.013738456765200958, 0.0, 0.0]
index = None

Aggregation Options

Assay aggregation options are modified with the wallaroo.assay_config.AssayBuilder.add_aggregation(aggregation: wallaroo.assay_config.Aggregation) method. The following options are provided:

• Aggregation.DENSITY (Default): Count the number/percentage of values that fall in each bin.
• Aggregation.CUMULATIVE: Empirical Cumulative Density Function style, which keeps a cumulative count of the values/percentages that fall in each bin.

The following example demonstrate the different results between the two.

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

#Aggregation.DENSITY - the default
assay_baseline.summarizer_builder.add_aggregation(wallaroo.assay_config.Aggregation.DENSITY)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show one analysis with the updated bins
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Density
metric = PSI
weighted = False
score = 0.027998899384779063
scores = [0.0, 0.00031071866653010913, 0.00030773024662236867, 0.002769365409558061, 0.015958199760662362, 0.008652885301406161, 0.0]
index = None

# Create the assay baseline
assay_baseline = wl.build_assay(assay_name="assays from date baseline", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# Set the assay parameters

# The end date to gather inference results
assay_baseline.add_run_until(datetime.datetime.now())

# Set the interval and window to one minute each, set the start date for gathering inference results
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

#Aggregation.CUMULATIVE
assay_baseline.summarizer_builder.add_aggregation(wallaroo.assay_config.Aggregation.CUMULATIVE)

# build the assay configuration
assay_config = assay_baseline.build()

# perform an interactive run and collect inference data
assay_results = assay_config.interactive_run()

# show one analysis with the updated bins
assay_results[0].chart()

baseline mean = 539895.63075
window mean = 520529.742265625
baseline median = 450283.78125
window median = 450867.6875
bin_mode = Quantile
aggregation = Cumulative
metric = PSI
weighted = False
score = 0.06899999999999992
scores = [0.0, 0.00799999999999998, 0.0, 0.02200000000000002, 0.03899999999999992, 0.0, 0.0]
index = None

Create Assay

With the assay previewed and configuration options determined, we officially create it by uploading it to the Wallaroo instance.

Once it is uploaded, the assay runs an analysis based on the window width, interval, and the other settings configured.

Assays are uploaded with the wallaroo.assay_config.upload() method. This uploads the assay into the Wallaroo database with the configurations applied and returns the assay id. Note that assay names must be unique across the Wallaroo instance; attempting to upload an assay with the same name as an existing one will return an error.

wallaroo.assay_config.upload() returns the assay id for the assay.

Typically we would just call wallaroo.assay_config.upload() after configuring the assay. For the example below, we will perform the complete configuration in one window to show all of the configuration steps at once before creating the assay.

# Build the assay, based on the start and end of our baseline time, 
# and tracking the output variable index 0
assay_baseline = wl.build_assay(assay_name="assays from date baseline tutorial samples", 
                                          pipeline=mainpipeline, 
                                          iopath="output variable 0",
                                          baseline_start=assay_baseline_start, 
                                          baseline_end=assay_baseline_end)

# set the width, interval, and assay start date and time
assay_baseline.window_builder().add_width(minutes=1).add_interval(minutes=1).add_start(assay_window_start)

# add other options
assay_baseline.summarizer_builder.add_aggregation(wallaroo.assay_config.Aggregation.CUMULATIVE)
assay_baseline.summarizer_builder.add_metric(wallaroo.assay_config.Metric.MAXDIFF)
assay_baseline.add_alert_threshold(0.5)

assay_id = assay_baseline.upload()

# wait 65 seconds for the first analysis run performed
time.sleep(65)

The assay is now visible through the Wallaroo UI by selecting the workspace, then the pipeline, then Insights.

Get Assay Info

Assay information is retrieved with the wallaroo.client.get_assay_info() which takes the following parameters.

This returns the following:

display(wl.get_assay_info(assay_id))

Get Assay Results

Once an assay is created the assay runs an analysis based on the window width, interval, and the other settings configured.

Assay results are retrieved with the wallaroo.client.get_assay_results method which takes the following parameters:

• IMPORTANT NOTE: This process requires that additional historical data is generated from the time the assay is created to when the results are available. To add additional inference data, use the Assay Test Data section above.

assay_results = wl.get_assay_results(assay_id=assay_id,
                     start=(datetime.datetime.now()-datetime.timedelta(days=2)),
                     end=datetime.datetime.now())

# Preview the assay analyses
assay_results.chart_scores()

assay_results[0].chart()

baseline mean = 519725.40603168664
window mean = 1116046.53125
baseline median = 448627.8125
window median = 1116046.53125
bin_mode = Quantile
aggregation = Cumulative
metric = MaxDiff
weighted = False
score = 0.8043912
scores = [0.0, 0.20159680638722555, 0.42115768463073855, 0.6027944111776448, 0.8043912175648703, 0.0, 0.0]
index = 4

List Assays

A list of assays is retrieved with the wallaroo.client.list_assays() method and takes the following parameters:

This returns a list of assays as filtered in reverse chronological order.

The errors for this method include:

• If the parameter workspace_id is not an integer.
• If the parameter workspace_name is not a String.

wl.list_assays()

Set Assay Active Status

Assays active status is either:

• True: The assay generates analyses based on the assay configuration.
• False: The assay will not generate new analyses.

Assays are set to active or not active with the wallaroo.client.set_assay_active which takes the following parameters.

First we will show the current active status.

In the following, set the assay status to False, then set the assay active status back to True.

display(wl.get_assay_info(assay_id))

Now we set the active status to False, and show the assay list to verify it is no longer active.

wl.set_assay_active(assay_id, False)
display(wl.get_assay_info(assay_id))

We resume the assay by setting it’s active status to True.

wl.set_assay_active(assay_id, True)
display(wl.get_assay_info(assay_id))

View Assay Filters

The following examples demonstrate various ways of listing assays and retrieving assay details, with filtering options.

List Assays with Filters

List all assays for workspaces the user is a member of.

wl.list_assays()

List all assays filtered by workspace id.

wl.list_assays(workspace_id=workspace_id)

List assays filtered by workspace name.

wl.list_assays(workspace_name=workspace_name)

Get Assay Results by Filter

Get assay results.

assay_results = wl.get_assay_results(assay_id=assay_id,
                     start=assay_window_start,
                     end=datetime.datetime.now())

assay_results.chart_scores()