Usage
h2o.glrm(training_frame, cols, k, model_id, validation_frame, loading_name, ignore_const_cols, transform = c("NONE", "DEMEAN", "DESCALE", "STANDARDIZE", "NORMALIZE"), loss = c("Quadratic", "L1", "Huber", "Poisson", "Hinge", "Logistic"), multi_loss = c("Categorical", "Ordinal"), loss_by_col = NULL, loss_by_col_idx = NULL, regularization_x = c("None", "Quadratic", "L2", "L1", "NonNegative", "OneSparse", "UnitOneSparse", "Simplex"), regularization_y = c("None", "Quadratic", "L2", "L1", "NonNegative", "OneSparse", "UnitOneSparse", "Simplex"), gamma_x = 0, gamma_y = 0, max_iterations = 1000, max_updates = 2 * max_iterations, init_step_size = 1, min_step_size = 0.001, init = c("Random", "PlusPlus", "SVD"), svd_method = c("GramSVD", "Power", "Randomized"), user_y = NULL, user_x = NULL, expand_user_y = TRUE, impute_original = FALSE, recover_svd = FALSE, seed, max_runtime_secs = 0)
Arguments
training_frame
An H2OFrame object containing the
variables in the model.
cols
(Optional) A vector containing the data columns on
which k-means operates.
k
The rank of the resulting decomposition. This must be
between 1 and the number of columns in the training frame, inclusive.
model_id
(Optional) The unique id assigned to the resulting model.
If none is given, an id will automatically be generated.
validation_frame
An H2OFrame object containing the
variables in the model.
loading_name
(Optional) The unique name assigned to the loading matrix X
in the XY decomposition. Automatically generated if none is provided.
ignore_const_cols
(Optional) A logical value indicating whether to ignore
constant columns in the training frame. A column is constant if all of its
non-missing values are the same value.
transform
A character string that indicates how the training data
should be transformed before running PCA. Possible values are "NONE":
for no transformation, "DEMEAN": for subtracting the mean of each
column, "DESCALE": for dividing by the standard deviation of each
column, "STANDARDIZE": for demeaning and descaling, and "NORMALIZE":
for demeaning and dividing each column by its range (max - min).
loss
A character string indicating the default loss function for numeric columns.
Possible values are "Quadratic" (default), "L1", "Huber", "Poisson", "Hinge"
and "Logistic".
multi_loss
A character string indicating the default loss function for enum columns.
Possible values are "Categorical" and "Ordinal".
loss_by_col
A vector of strings indicating the loss function for specific
columns by corresponding index in loss_by_col_idx. Will override loss for
numeric columns and multi_loss for enum columns.
loss_by_col_idx
A vector of column indices to which the corresponding loss
functions in loss_by_col are assigned. Must be zero indexed.
regularization_x
A character string indicating the regularization function for
the X matrix. Possible values are "None" (default), "Quadratic", "L2", "L1",
"NonNegative", "OneSparse", "UnitOneSparse", and "Simplex".
regularization_y
A character string indicating the regularization function for
the Y matrix. Possible values are "None" (default), "Quadratic", "L2", "L1",
"NonNegative", "OneSparse", "UnitOneSparse", and "Simplex".
gamma_x
The weight on the X matrix regularization term.
gamma_y
The weight on the Y matrix regularization term.
max_iterations
The maximum number of iterations to run the optimization loop.
Each iteration consists of an update of the X matrix, followed by an update
of the Y matrix.
max_updates
The maximum number of updates of X or Y to run. Each update consists
of an update of either the X matrix or the Y matrix. For example, if max_updates = 1
and max_iterations = 1, the algorithm will initialize X and Y, update X once, and
terminate without updating Y.
init_step_size
Initial step size. Divided by number of columns in the training
frame when calculating the proximal gradient update. The algorithm begins at
init_step_size and decreases the step size at each iteration until a
termination condition is reached.
min_step_size
Minimum step size upon which the algorithm is terminated.
init
A character string indicating how to select the initial Y matrix.
Possible values are "Random": for initialization to a random array from the
standard normal distribution, "PlusPlus": for initialization using the clusters
from k-means++ initialization, or "SVD": for initialization using the
first k right singular vectors. Additionally, the user may specify the
initial Y as a matrix, data.frame, H2OFrame, or list of vectors.
svd_method
(Optional) A character string that indicates how SVD should be
calculated during initialization. Possible values are "GramSVD": distributed
computation of the Gram matrix followed by a local SVD using the JAMA package,
"Power": computation of the SVD using the power iteration method, "Randomized":
(default) approximate SVD by projecting onto a random subspace (see references).
user_y
(Optional) A matrix, data.frame, H2OFrame, or list of vectors specifying the
initial Y. Only used when init = "User". The number of rows must equal k.
user_x
(Optional) A matrix, data.frame, H2OFrame, or list of vectors specifying the
initial X. Only used when init = "User". The number of columns must equal k.
expand_user_y
A logical value indicating whether the categorical columns of user_y
should be one-hot expanded. Only used when init = "User" and user_y is specified.
impute_original
A logical value indicating whether to reconstruct the original training
data by reversing the transformation during prediction. Model metrics are calculated
with respect to the original data.
recover_svd
A logical value indicating whether the singular values and eigenvectors
should be recovered during post-processing of the generalized low rank decomposition.
seed
(Optional) Random seed used to initialize the X and Y matrices.
max_runtime_secs
Maximum allowed runtime in seconds for model training. Use 0 to disable.