---
alias: user-guide-automation-task-dataplatform-mlprediction
tags:
- automation tasks
description: "This task calculates ML predictions using an ML Model and IoT event data"
---

# ML Prediction

:core-entitytypes-mlmodel-lg:{style="font-size: 100px"}

The task `ML Prediction` is a Task that calculates a prediction based on an ML Model and returns its result.

![mlprediction_01][mlprediction_01]

## Details

| Fields            | Details                                                                            |
| ----------------- | ---------------------------------------------------------------------------------- |
| Scope             | Connect IoT <br> Factory Automation <br> Enterprise Integration <br> Data Platform |
| Workflow          | Data flow <br> Control flow                                                        |
| `IsProtocol` flag | :material-close:{ style="color: #EF5350" }                                       |

## Settings

On the `General` Tab, you have the usual settings and the following options settings:

| Name          | Data Type | Default  | Description                                                  |
| ------------- | --------- | -------- | ------------------------------------------------------------ |
| ML Model | `ML Model` | | **ML Model** entity from MES |
| IoT Event Definition | `IoT Event Definition` | |**IoT Event Definition** entity from MES |
| Endpoint| String | `http://mlplatformagent:8080` | **ML Model** Prediction Endpoint |

![mlprediction_02][mlprediction_02]

If an IoT Event Definition is selected, the `Inputs` Tab will be hidden and the `Features` Tab will show up. In there you need to map the IoT Event Definition Properties to the ML Model Features:

![mlprediction_03][mlprediction_03]

## Inputs

The inputs depend on the ML Model that was selected and on whether or not an IoT Event Definition was selected.

| Name         | Data Type | Description                                  |
| ------------ | --------- | ------------------------------------------ |
| iotEvent     | `string`  | The whole event. Appears when an IoT Event Definition was selected                 |
| eventId      | `string`  | The Id of the Event. Appears when no IoT Event Definition was selected                 |
| FeatureNames | `string`  | An input for each Feature of the ML Model. Appears when no IoT Event Definition was selected |
| Activate     | `Any`     | Activates the task execution                 |

If no IoT Event Definition was selected, there will be an input for the Event Id and an input for each Feature:

![mlprediction_04][mlprediction_04]

If an IoT Event Definition was selected, there will be an input for the whole Event:

![mlprediction_05][mlprediction_05]

## Outputs

| Name     | Data Type | Description                                                  |
| -------- | --------- | ---------------------------------------------------------- |
| result   | `any` | The full result of the prediction. Includes info about the ML Model version that was used  |
|prediction| `any` | The prediction itself. The data type depends on the ML Model Type       |
| score  | `Decimal` | The score of the prediction. This might be hidden depending on the ML Model Type                         |
| Success  | `Boolean` | If there was no problem emits a true                         |
| Error    | `Error`   | Error that occurred during the processing of this task       |

## Behavior

The task is used to calculate a prediction based on an **ML Model** when an **IoT Event** is received.

## Remarks

This tasks has **2 modes** depending on whether or not an **IoT Event Definition** is selected in the settings.

[mlprediction_01]: ./images/mlprediction_01.png
[mlprediction_02]: ./images/mlprediction_02.png
[mlprediction_03]: ./images/mlprediction_03.png
[mlprediction_04]: ./images/mlprediction_04.png
[mlprediction_05]: ./images/mlprediction_05.png