# Root cause analysis

> When analyzing a business process, you may want to determine which fields are most associated with a certain outcome. This should help you to act on the root causes associated with the outcome. For example, in the Purchase-to-Pay process, you may want to analyze the influence of purchase orders that have maverick buying tags assigned.

## What is root cause analysis?

When analyzing a business process, you may want to determine which fields are most associated with a certain outcome. This should help you to act on the root causes associated with the outcome. For example, in the Purchase-to-Pay process, you may want to analyze the influence of purchase orders that have maverick buying tags assigned.

With **Root cause analysis**, you can compare the influence of case fields on a certain behavior to find significant data influencers for specific process situations. A set of cases is defined based on the period filter. This selection is called *Reference cases*. Within this set of cases, you can select the behavior that you want to analyze. For example, cases with a certain tag. This selection is called the *Selected cases*. The influence of a field is based on the number of occurrences in the selected cases.

## Root cause analysis dashboard

Use the **Root cause analysis** dashboard to compare the influence of case properties on a set of selected cases within a reference set of cases.

## Performing a root cause analysis

Follow these steps to perform a root cause analysis.

| Step | Action |
| --- | --- |
| 1 | Use the **Timeframe** filter to define the set of *Reference cases*. |
| 2 | Select **Root cause analysis** in the menu on the left of the dashboard. |
| 3 | Use the **Filter** panel to create filters that define the set of *Selected cases*, which are the cases you want to analyze the influence on. |
| 4 | Select the field you want to use for your analysis from the selector. |

:::note
Up to 1000 results are visible. When there is too much data, a warning icon is displayed.
:::

### Node limit slider

The **Node limit** slider enables you to reduce the complexity of the Root Cause Analysis tree, which increases the readability of the graph. By default, the detail of the Root Cause Analysis is automatically determined. You can use the **Node limit** slider to change the number of nodes shown.

### Zoom in/zoom out

You can use the zoom in/zoom out buttons at the bottom to change the magnification of the Root Cause Analysis tree. The following table describes the buttons.

| Button | Click to... |
| --- | --- |
| ![Zoom in](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-zoom-in-Zoom_in_icon-52f2f59b-d39108c8.png) | Zoom in |
| ![Zoom out](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-zoom-out-Zoom_out_icon-c829a49c-57714c93.png) | Zoom out |
| ![Reset to default view](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-reset-to-default-view-Reset_zoom_icon-4c43d276-915c13f3.png) | Reset to the default view |

:::note
You can also use the mouse wheel to zoom in or zoom out.
:::

### Influence

The **Root cause analysis** tree displays the **value (%)**, the number of occurrences in the **Selected cases**, and the number of occurrences in the **Reference cases** for the field selected in the dashboard. A large deviation from the **Reference cases** indicates a possible high influence on the selection.

The above image shows that Maverick buying for example occurs less in the **2800 - BestRun China** company (-2%) than in other companies in the reference data, and that Maverick buying occurs more in the **5000 - BestRun Japan 5000** company (10%) than in other companies in the reference data.

:::note
The **value (%)** in the start node is the *global baseline percentage*, whereas the **value (%)** in the other nodes is the *Influence (%)* which represents the *deviation* of the node’s selected percentage from the *global baseline percentage*.
:::

#### Show significant influencers option

The **Show significant influencers** option enables you to zoom in by displaying the cases with a statistical significant influence. This should help you identify the cases that have the most impact on the selection. This statistical significance is computed from both the *Influence (%)* and the amount of cases a certain field has.

![docs image](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-show-significant-influencers-option-244759-857e2d76-45cdf25f.webp)

## Adding layers

If desired, you can add more layers to the **Root cause analysis**.

![Add layer option](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-add-layer-option-15569-7594a04f-f057c740.webp)

![Root cause analysis example](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-root-cause-analysis-example-304110-3a32c479-f1ed7e0f.webp)

In the above example, the combination of fields results in a set of *Selected cases* that has not enough (relevant) data to determine influencers. In this case, you can narrow down the set of *Reference cases* by adding a filter on the dashboard.

![Adding a filter on the Root cause analysis dashboard](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-adding-a-filter-on-the-root-cause-analysis-dashboard-244778-a4b3675d-8ce4f415.webp)

The following illustration shows the result.

![Result after adding a filter](https://dev-assets.cms.uipath.com/assets/images/process-mining/process-mining-result-after-adding-a-filter-304114-edde9ddd-12850d63.webp)

When hovering over the fields in the tree, the **Influence (%)**, the **Reference cases**, and the **Selected cases** are displayed.

The following table describes the metrics.

| Metric | Description |
| --- | --- |
| Influence (%) | The deviation of the *Selected cases* from the *Reference cases*. |
| Selected cases | The number of cases for the field in the total set of *Selected cases*. |
| Reference cases | The number of cases for the field in the total set of *Reference cases*. |