Data prediction operations

This category contains all operations related to the Data prediction feature.

Regression

Creates and trains the Linear Regression Model on the specified database. This operation has the following properties.

Property name Type Description
Description String Displayed name of the operation. This property is only used in the HMI and has no effect at run-time.
Database name String Name of the database table. This property supports usage of internal bot variables.
Model name String Name of the ML model file. This property supports usage of internal bot variables.
Key for accuracy String Unique reference in memory to the trained model accuracy. This property supports usage of internal bot variables.
Column to predict String Name of the database column containing the expected result for training the ML model. This property supports usage of internal bot variables.
Columns to parse Set Names of database columns containing the input features for training the ML model. Only specified columns and the “column to predict” will be used in model training. Column names support usage of internal bot variables.
Condition for rows to parse C# lambda expression

C# lambda expression – function taking the database rows as an input and filtering which rows to use in training. This property supports usage of internal bot variables.

As an example, the following string can be used to train the model on all rows having the value of ABC property less than 30. x => x.ABC < 30

The following string can be used to train the model on all rows having the value of ABCD property equals to "test". x => x["ABCD"] == "test"
Algorithm Enum Algorithm used for model training.
Optimization metric Enum Algorithm metric to be optimized during the model training. The result metric value (accuracy) will be saved in the bot’s internal variable defined by the “Key for accuracy” key.

Predict

Apply the previously trained ML model on the specified data row. This operation has the following properties.

Property name Type Description
Description String Displayed name of the operation. This property is only used in the HMI and has no effect at run-time.
Model name String Name of the ML model file. This property supports usage of internal bot variables.
Result key String Unique reference in memory to store the prediction result. This property supports usage of internal bot variables.
Scope Enum

Accessibility scope of the result in memory specified by the "Result key" reference. The following values are available.

  • Private – value is available only for the current bot

  • Internal – value is available for the current thread, i.e. for parent bots, for the current bot and for all children executed bots.

  • Shared – value is available for all bots.
Row’s key String Unique reference in memory to database row to apply the model on. This property supports usage of internal bot variables.
Sample properties Set A set of table columns and corresponding values to use as the ML model input. If some column was not used in the model training, it will be ignored. Both properties and values supports usage of internal bot variables.