tLinearRegressionModel properties for Apache Spark Batch

Schema and Edit schema

A schema is a row description. It defines the number of fields (columns) to be processed and passed on to the next component. When you create a Spark Job, avoid the reserved word line when naming the fields.

Click Edit schema to make changes to the schema. If the current schema is of the Repository type, three options are available:

• View schema: choose this option to view the schema only.

• Change to built-in property: choose this option to change the schema to Built-in for local changes.

• Update repository connection: choose this option to change the schema stored in the repository and decide whether to propagate the changes to all the Jobs upon completion. If you just want to propagate the changes to the current Job, you can select No upon completion and choose this schema metadata again in the Repository Content window.

Label column

Select the input column used to provide Double-type labels (values of the dependent variable in terms of linear regression). The records of this column are used as the potential situations (the variation of the dependent variable in terms of linear regression) a given element could fall into.

Feature column

Select the input column used to provide Vector-type features (values of the independent or explanatory variable in terms of linear regression). Very often, this column is the output of the feature engineering computations performed by tModelEncoder.

Save the model on file system

Select this check box to store the model in a given file system. Otherwise, the model is stored in memory. The button for browsing does not work with the Spark Local mode; if you are using the Spark Yarn or the Spark Standalone mode, ensure that you have properly configured the connection in a configuration component in the same Job, such as tHDFSConfiguration.

ElasticNet mixing parameter

Enter the ElasticNet coefficient (numerical value) used for the regularization calculation in order to control the bias/variance trade-off in feature selection. ElasticNet is the combination of L1 regularization and L2 regularization.

The value to be put varies between 0.0 and 1.0, indicating the weights of the L1 regularization and the L2 regularization in the ElasticNet combination. When the value is 0.0, the regularization is actually equivalent to the L2 regularization; when the value is 1.0, it is equivalent to the L1 regularization.

For further information about how ElasticNet is implemented in Spark, see ML linear methods, in which the related formula shows how the value you put (α in that formula) is used to calculate the ElasticNet regularization.

For further information about ElasticNet, see Regularization and variable selection via the elastic net.

Fit an intercept term

Select this check box to allow the tLinearRegressionModel to automatically calculate the intercept constants and include them in the regression computation.

In general, intercept should be present to guarantee that the residuals of your model have a mean of zero.

Standardize features before fitting model

Select this check box to scale the features to make them normally distributed.

Maximum number of iterations

Enter the number of iterations you want the Job to perform to train the model.

Regularization

Enter the regularization coefficient (numerical value) to be used along with ElasticNet for the regularization calculation.

For further information about how this parameter is implemented in Spark, see ML linear methods, in which the related formula shows how the value you put (λ in that formula) is used to calculate the eventual regularization.

Convergence tolerance

Enter the convergence score which the iterations are expected to obtain.

In general, smaller value will result in higher accuracy in the prediction at the cost of more iterations.

But note that in some cases, your model may not be able to reach the convergence you put despite of whatever number of iterations you want the Job to perform. This failure to converge might indicate that the convergence score you use is not realistic to the features you are processing and therefore, you need to process these features to a greater degree.

Solver algorithm

Select the algorithm used for optimization.

• Normal: this algorithm uses normal equations.

• L-BFGS: this algorithm approximates the BFGS algorithm using a limited amount of computer memory.

• Auto: the component select either of the above-mentioned algorithms.

Usage rule

This component is used as an end component and requires an input link.

You can accelerate the training process by adjusting the stopping conditions such as the maximum number of iterations or the convergence tolerance but note that the training that stops too early could impact its performance.

Model evaluation

The parameters you need to set are free parameters and so their values may be provided by previous experiments, empirical guesses or the like. They do not have any optimal values applicable for all datasets.

Therefore, you need to train the relationship model you are generating with different sets of parameter values until you can obtain the best evaluation result. But note that you need to write the evaluation code yourself to rank your model with scores.

For general information about validating a regression-based relationship model, see https://en.wikipedia.org/wiki/Regression_validation.

Spark Connection

This connection is effective on a per-Job basis.