> For the complete documentation index, see [llms.txt](https://docs.coherent.global/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.coherent.global/navigation/testing-center/testbed-details.md).
# Testbed details
In the **Testing Center**, a Spark service can be applied against a dataset to determine the calculated outputs for many records at a time. The main use cases for this feature are for regression testing, model simulation, and portfolio projection use cases.
After choosing a specific service from the [Folder overview](/navigation/folder-overview.md), select the  icon from the service sidebar. From the [Testbeds](/navigation/testing-center/testbeds.md), select a **Testbed name** to view the **Testbed details**.
## Testing Center terminology
* A test case is a single test case record. A test case can be composed of multiple single value and table inputs.
* A testbed contains multiple test cases.
* After a testbed is run against a Spark service, the output is a testbed result. It should contain the same number of records as the testbed.
## Testbed information
The top of the **Testbed details** page shows information about the testbed added, modified, last run, version, and test cases.
## Add test cases
Add test cases to an existing testbed and increment the testbed version.
1. Select **Add test cases**.
2. **Need the template file to add your test cases?** Select **Download it here**.
3. Open the downloaded Excel file. The file includes a table with the input names as headings.
4. Follow the instructions in the file to add additional test cases.
5. Select **Browse** or drag the workbook into the upload area. Spark supports `.xlsx` and `.xlsm` files.
6. Spark will show the number of existing test cases and number of new test cases.
7. You can enter a new **Testbed description**.
8. Confirm the change by selecting **Add**.
9. The testbed will have a minor [semantic version](https://semver.org/) update, for example `1.2.0` to `1.3.0` and the test case count will increase.
If the new test cases already exist in the testbed, they will be removed and noted in the duplicated test case count.
## Run testbed
Select ▶️ **Run** to run the testbed.
1. Select the **Service version** that you want to run the testbed against.
2. **Select service type**: Stay with the default selection of **Neuron**.
3. **Select execution mode**: **Legacy** execution is the current method of running tests. **Batch** execution mode is experimental - it should require less configuration for batch size and should process in a shorter period of time.
4. Enter a **Testbed result name**.
5. Optionally, enter a **Testbed result description**.
6. If your service has [Sub functions with subservices](/build-spark-services/additional-mappings/sub-functions-with-subservices.md), you can select which subservices to include.
7. Select **Run**.
1. You can monitor the progress in the modal.
2. Alternatively, you can **Close** the modal and monitor progress in **Background activity**. Once it is completed select the **Run testbeds** activity.
8. The **Run testbed** modal should indicate **This testbed was successfully run.** with some statistics about the testbed result.
9. To download the result immediately, select **Download result**. Follow the instructions in [#download-testbed-result](#download-testbed-result "mention").
{% hint style="warning" %}
If you encounter an error relating to timeout and batch size, we suggest the following actions:
* Edit the [Service Documentation](/navigation/service-documentation.md#version-details) **Batch Size Override** to a lower number. This reduces the number of test cases that are submitted at a time which reduces the likelihood of a timeout as fewer records will be processed together. The default batch size value is `100`.
* If you frequently encounter this across many Spark services, the batch size value can be adjusted for new services. Contact [Support](/support/faq.md).
{% endhint %}
## Testbed actions
Additional actions can be performed on a testbed from the ellipsis menu.
See [Testbeds](/navigation/testing-center/testbeds.md#testbed-actions).
## Change how testbed results are displayed
* **Search**: Enter a search term and the list will update.
* **Source**: Filter for a particular testbed result source.
| Source | Description |
| ---------------------- | ----------------------------------------------------------------------------------------------------------------- |
| **Spark** | Testbed result was run from the Testing Center. |
| **Coherent Assistant** | Testbed result was created by [Coherent Assistant](https://docs.coherent.global/assistant/) **Testbeds** feature. |
| **User upload** | Testbed result was uploaded by any user. |
* **Service version**: Filter for a specific service version.
* **Sort**: Change the sort order by selecting the **List view** headers.
* Select the expand for a testbed result to view the **Subservices**, **Service type**, **Validation errors**, **Output errors**, **Run time**, **Calculation time**, **Description**.
## Compare testbed results
Download a report that will indicate where there are mismatches between two sets of testbed results. The main use cases for this feature are for regression testing and analysis of model changes between service versions.
1. Select two testbed results.
2. Select **Compare results**.
3. In the modal, select the comparison options.
4. **Choose download format**. `CSV` is recommended as the most compact data format.
| Format | Description |
| --------------------- | ----------------------------------------------------------------------------------------------------------- |
| **Excel (fast beta)** | This is an experimental feature. This uses an alternative faster method to write a testbed result to Excel. |
| **Excel** | Return the comparison in a single Excel file. |
5. **Choose whether to return the full comparison or mismatches only.** If you are performing regression testing against an approved set of results, selecting **Mismatches only** will reduce the file generation time and size.
6. **Choose whether to return the inputs with the testbed comparison.** Excluding the inputs will result in smaller file sizes and faster downloads.
7. **Choose how you would like headings to be ordered**. **Data order** sorts the mapped inputs and outputs in the order they appear in the workbook.
8. **Select particular outputs to compare**. By default, all outputs are included. Choosing a subset of the outputs makes it easier to review the results and will result in smaller file sizes and faster downloads.
9. Select **Download**.
10. Spark will show the mismatch results.
11. The download task initiates and can be monitored in **Background activity**.
12. Once the download is ready, it will automatically start to download.
13. If you missed the download, the download can be initiated again in **Background activity**.
## Upload testbed result
To enable broader comparisons across testbeds, you can upload a testbed result to Spark and use it to [#compare-testbed-results](#compare-testbed-results "mention").
1. Select **Upload testbed result**.
2. **Need the latest version of this testbed to add your testbed result?** Select **Download it here**.
3. Open the downloaded Excel file. The file includes a table with the input data and output names as headings.
4. Follow the instructions in the file to populate the outputs for each test case.
5. Select **Browse** or drag the workbook into the upload area. Spark supports `.xlsx` and `.xlsm` files.
6. Enter a **Testbed result name**.
7. Optionally, enter a **Testbed result description**.
8. Spark will show the number of **Test cases with results**.
9. Select **Update**.
10. The new testbed result is added to the list with the **Source** **User Upload**.
## Testbed result actions
Additional actions can be performed on a testbed result from the ellipsis menu.
### Download testbed result
1. In the modal, choose the desired download options.
2. **Choose download format**. `CSV` is recommended as the most compact data format.
| Format | Description |
| --------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **CSV** | Return the testbed result in a `ZIP` file containing `CSV` files. Files are chunked at `25 MB`. |
| **CSV expanded tables** | **CSV** plus **expanded tables** will write table inputs and outputs to separate CSV files making it easy to consume testbed result that include tables. |
| **Excel** | Return the testbed result in a single Excel file. |
| **Excel (expanded tables)** | Excel plus expanded tables will write table inputs and outputs to separate worksheets making it easy to consume testbed result that include tables. This feature will not work if the testbed contains:
- A large number of test cases that include input or output tables.
- Very large input or output tables.
|
| **Excel (fast beta)** | A faster version of **Excel (expanded tables)**. |
3. **Choose whether to return the inputs with the testbed result.** Excluding the inputs will result in smaller file sizes and faster downloads.
4. **Choose how you would like headings to be ordered**. **Data order** sorts the mapped inputs and outputs in the order they appear in the workbook.
5. **Select particular outputs to download**. By default, all outputs are included. Choosing a subset of the outputs makes it easier to review the results and will result in smaller file sizes and faster downloads.
6. Select **Download**.
7. The download task initiates and can be monitored in **Background activity**.
8. Once the download is ready, it will automatically start to download.
9. If you missed the download, the download can be initiated again in **Background activity**.
### Delete testbed result
Before a testbed result is deleted, Spark will request confirmation for the delete action. After confirmation, everything related to the testbed result will be deleted. There is no process to recover deleted testbed results.
### Explore test case in API Tester
View the inputs for a particular test case in the [API Tester](/navigation/api-tester.md). Verify the calculations for the particular use case.
1. From a downloaded testbed result file, identify the test case record you want to examine and find its **UUID** value.
2. Enter the test case UUID.
3. Select **Submit**.
4. The [API Tester](/navigation/api-tester.md) appears with the input data from the test case.
### Aggregate testbed result
If you have a large testbed but are only interested in summaries of the output data, Spark can apply the summation directly.
The interface is designed to allow you to define all of your aggregations together at once.
* The aggregation list pane on the left lists all the aggregation blocks.
* Aggregation blocks are used to define aggregation operations. Within an aggregation block, multiple aggregation operations can be defined against the same filters and group by.
1. We start with the default aggregation block named **Aggregation #1**.
2. If you want to filter the data before the aggregation, select **+** for **Filters**.
1. Select whether to filter for an **Input** or **Output** field. Then select the **Field name**.
2. For tables, you must also select the column(s) to aggregate. Any columns that are not selected will copy their table entries from the first test case.
3. Select the filter **Operator** and enter a **Value**.
4. You can apply a compound filter by adding another filter condition.
5. To remove a filter, select the `-` button.
3. Select the aggregation(s) you want to perform.
1. Under **Aggregations**, select the **Output name**, and the aggregation **Operator**.
2. Add additional aggregation operations.
3. To remove an aggregation, select the `-` button.
4. If you want to group the by any input or output variables, select **+** for **Group By**.
1. Select whether to group by an **Input** or **Output** field. Then select the **Field name**.
2. You can apply a compound group by by adding another group by field.
3. To remove a group by, select the `-` button.
5. You can cancel the aggregation by selecting the **Cancel** button.
6. If you are ok with the aggregation, select the **Save** button.
7. The aggregation block name is renamed and the filter, aggregation, and group by fields are no longer editable.
8. You can add additional aggregation blocks by selecting **New aggregation**.
9. After you have completed the aggregation blocks, select **Submit**.
10. The aggregation task initiates and can be monitored in **Background activity**.
11. Once the download is ready, it will automatically start to download.
12. If you missed the download, the download can be initiated again in **Background activity**.
Each aggregation operation that is defined will have a separate worksheet in the Excel.