H2O Generalized Low Rank Models (Missing Value Impute)

Apply a Generalized Low Rank Model (GLRM) using H2O to reconstruct missing values or identify important features in a dataset. Note that if the input data contains no missing values, the reconstructed data returned by this node will be the same as the input data.

Options

General Settings

Column selection

Select columns for the input data.

Ignore constant columns

Select to ignore constant columns.

Use static random seed

Select to use static seed for randomization.

Algorithm Settings

Transformation

Specify the transformation method for the training data (None, Standardize, Normalize, Demean, or Descale). The default is None.

Rank of matrix approximation

Specify the rank of matrix approximation (required).

Numerical loss function

Specify the numeric loss function (Quadratic, Absolute, Huber, Poisson, or Periodic).

Length of period

Specify the length of the period (only enabled when the numerical loss function is set to Periodic).

Categorical loss function

Specify the categorical loss function (Categorical or Ordinal).

Regularization function for X matrix

Specify the regularization function for the X matrix (None, Quadratic, L2, L1, NonNegative, OneSparse, UnitOneSparse, or Simplex).

Regularization function for Y matrix

Specify the regularization function for the Y matrix (None, Quadratic, L2, L1, NonNegative, OneSparse, UnitOneSparse, or Simplex).

Regularization weight on the X matrix

Specify the regularization weight on the X matrix.

Regularization weight on the Y matrix

Specify the regularization weight on the Y matrix.

Max number of iterations

Specify the maximum number of training iterations. The maximum value is 1000000 (max_iterations).

Max number of updates

Specify the maximum number of updates.

Initial step size

Specify the initial step size.

Min step size

Specify the minimum step size. This value should be between 0 and the initial step size.

Initialization mode

Specify the initialization mode (Random, SVD, PlusPlus, or User).

SVD method

Specify the method for computing SVD during initialization (GramSVD, Power, Randomized). Note that Power and Randomized are currently experimental.

Max runtime in seconds

Specify the maximum allowed runtime in seconds for model training (max_runtime_secs).

Views

GLRM model summary

A table with the GLRM model summary

Workflows

• test-real-life-edge-casesKNIME Hub

Links

• No links available

Developers