Machine Learning
This context is a container, holding all trainable units for a particular user.
Unique Actions
Create
This action is used to create a new trainable unit. It allows the user to specify the basic properties for the new trainable unit and configure it immediately after creation.
Action Type: | |
Permissions: | Accessible at Manager permission level |
Common Actions
Create From Template, Replicate To Children, Import, Export, Edit Context Permissions, Monitor Related Events, Search/Filter, various Grouped Actions according to child contexts.
Context States and Icons
This context has no states. It is always represented by the 
icon.
Advanced Information |
Context Information
Context Type: machineLearning
Context Name: machineLearning
Context Description: Machine Learning
Context Path: users.USER_NAME.machineLearning
Context Mask: users.*.machineLearning
Context Permissions
Level | Description |
None | No access allowed. |
Observer | Basic event monitoring. |
Operator | Same as Observer. |
Manager | Trainable unit creation, export and import. |
Engineer | Same as Manager. |
Administrator | Same as Manager. |
Public Variables (Properties)
This context has no public variables (properties).
Public Functions
Common Functions: makeCopy (Make Copy), delete (Delete)
Create
Creates a new trainable unit.
Function Name: | create |
Permissions: | Accessible at Manager permission level |
Input Records: | 1 |
Input Format: | Same as format of childInfo variable in Trainable Unit context. |
Output Records: | 0 |
Output Format: | none |
Exponential Smoothing
This function performs Holt-Winters triple exponential smoothing.
Function Name: | exponentialSmoothing | ||
Permissions: | Accessible at Observer permission level | ||
Input Records: | 1 | ||
Input Format: | |||
Name | Type | Description | |
|---|---|---|---|
data | Data Table | Table containing at least one numeric field (i.e. label column) with time series data records | |
labelFieldName | String | Name of the label column (can be null if the Data Table at the "data" field has only one field) 1 - 50 characters Nullable | |
valueSmoothingFactor | Integer | Smoothing factor for the series values (must be between 0 and 1) | |
trendSmoothingFactor | Integer | Smoothing factor for the trend (must be between 0 and 1) | |
seasonalSmoothingFactor | Integer | Smoothing factor for the seasonal component (must be between 0 and 1) | |
seasonCycleLength | Integer | Length of the seasonal cycle (e.g. 12 for monthly data; 4 for quarterly data) | |
numPredictedValues | Integer | Number of values to be predicted | |
Output Records: | 0...unlimited | ||
Output Format: | Dynamic, contains all fields from the Data Table at the "data" input field plus a field containing smoothed data and predicted values (if numPredictedValues > 0): | ||
Name | Type | Description | |
predictedValue | Double | Nullable | |
 | Note that the function requires 2*numPredictedValues + 1 data points for initialisation in order to calculate smoothed (as well as predicted) values. |
Public Events
Common Events: info (Information)
Was this page helpful?