public static final class GLMV3.GLMParametersV3 extends water.api.schemas3.ModelParametersSchemaV3<GLMModel.GLMParameters,GLMV3.GLMParametersV3>
auc_type, categorical_encoding, checkpoint, custom_distribution_func, custom_metric_func, distribution, export_checkpoints_dir, fold_assignment, fold_column, gainslift_bins, huber_alpha, ignore_const_cols, ignored_columns, keep_cross_validation_fold_assignment, keep_cross_validation_models, keep_cross_validation_predictions, max_categorical_levels, max_runtime_secs, model_id, nfolds, offset_column, parallelize_cross_validation, quantile_alpha, response_column, score_each_iteration, stopping_metric, stopping_rounds, stopping_tolerance, training_frame, tweedie_power, validation_frame, weights_column| Constructor and Description |
|---|
GLMParametersV3() |
append_field_arrays, extractDeclaredApiParameters, fields, fillFromImpl, fillImpl, getAdditionalParameters, writeParametersJSONcreateAndFillImpl, createImpl, extractVersionFromSchemaName, fillFromAny, fillFromBody, fillFromImpl, fillFromImpl, fillFromParms, fillFromParms, fillFromParms, fillImpl, getImplClass, getImplClass, getSchemaName, getSchemaType, getSchemaVersion, init_meta, markdown, markdown, newInstance, newInstance, setField, setSchemaType_doNotCallpublic static final java.lang.String[] fields
@API(help="Seed for pseudo random number generator (if applicable).",
gridable=true)
public long seed
@API(help="Family. Use binomial for classification with logistic regression, others are for regression problems.",
values={"AUTO","gaussian","binomial","fractionalbinomial","quasibinomial","ordinal","multinomial","poisson","gamma","tweedie","negativebinomial"},
level=critical)
public GLMModel.GLMParameters.Family family
@API(help="Tweedie variance power",
level=critical,
gridable=true)
public double tweedie_variance_power
@API(help="Dispersion learning rate is only valid for tweedie family dispersion parameter estimation using ml. It must be > 0. This controls how much the dispersion parameter estimate is to be changed when the calculated loglikelihood actually decreases with the new dispersion. In this case, instead of setting new dispersion = dispersion + change, we set new dispersion = dispersion + dispersion_learning_rate * change. Defaults to 0.5.",
level=expert,
gridable=true)
public double dispersion_learning_rate
@API(help="Tweedie link power.",
level=critical,
gridable=true)
public double tweedie_link_power
@API(help="Theta",
level=critical,
gridable=true)
public double theta
@API(help="AUTO will set the solver based on given data and the other parameters. IRLSM is fast on on problems with small number of predictors and for lambda-search with L1 penalty, L_BFGS scales better for datasets with many columns.",
values={"AUTO","IRLSM","L_BFGS","COORDINATE_DESCENT_NAIVE","COORDINATE_DESCENT","GRADIENT_DESCENT_LH","GRADIENT_DESCENT_SQERR"},
level=critical)
public GLMModel.GLMParameters.Solver solver
@API(help="Distribution of regularization between the L1 (Lasso) and L2 (Ridge) penalties. A value of 1 for alpha represents Lasso regression, a value of 0 produces Ridge regression, and anything in between specifies the amount of mixing between the two. Default value of alpha is 0 when SOLVER = \'L-BFGS\'; 0.5 otherwise.",
level=critical,
gridable=true)
public double[] alpha
@API(help="Regularization strength",
required=false,
level=critical,
gridable=true)
public double[] lambda
@API(help="Use lambda search starting at lambda max, given lambda is then interpreted as lambda min.",
level=critical)
public boolean lambda_search
@API(help="Stop early when there is no more relative improvement on train or validation (if provided).") public boolean early_stopping
@API(help="Number of lambdas to be used in a search. Default indicates: If alpha is zero, with lambda search set to True, the value of nlamdas is set to 30 (fewer lambdas are needed for ridge regression) otherwise it is set to 100.",
level=critical)
public int nlambdas
@API(help="Perform scoring for every score_iteration_interval iterations.",
level=secondary)
public int score_iteration_interval
@API(help="Standardize numeric columns to have zero mean and unit variance.",
level=critical,
gridable=true)
public boolean standardize
@API(help="Only applicable to multiple alpha/lambda values. If false, build the next model for next set of alpha/lambda values starting from the values provided by current model. If true will start GLM model from scratch.",
level=critical)
public boolean cold_start
@API(help="Handling of missing values. Either MeanImputation, Skip or PlugValues.",
values={"MeanImputation","Skip","PlugValues"},
level=expert,
direction=INOUT,
gridable=true)
public GLMModel.GLMParameters.MissingValuesHandling missing_values_handling
@API(help="If set to dfbetas will calculate the difference in beta when a datarow is included and excluded in the dataset.",
values="dfbetas",
level=expert,
gridable=false)
public GLMModel.GLMParameters.Influence influence
@API(help="Plug Values (a single row frame containing values that will be used to impute missing values of the training/validation frame, use with conjunction missing_values_handling = PlugValues).",
direction=INPUT)
public water.api.schemas3.KeyV3.FrameKeyV3 plug_values
@API(help="Restrict coefficients (not intercept) to be non-negative.") public boolean non_negative
@API(help="Maximum number of iterations. Value should >=1. A value of 0 is only set when only the model coefficient names and model coefficient dimensions are needed.",
level=secondary)
public int max_iterations
@API(help="Converge if beta changes less (using L-infinity norm) than beta esilon. ONLY applies to IRLSM solver.",
level=expert)
public double beta_epsilon
@API(help="Converge if objective value changes less than this. Default (of -1.0) indicates: If lambda_search is set to True the value of objective_epsilon is set to .0001. If the lambda_search is set to False and lambda is equal to zero, the value of objective_epsilon is set to .000001, for any other value of lambda the default value of objective_epsilon is set to .0001.",
level=expert)
public double objective_epsilon
@API(help="Converge if objective changes less (using L-infinity norm) than this, ONLY applies to L-BFGS solver. Default (of -1.0) indicates: If lambda_search is set to False and lambda is equal to zero, the default value of gradient_epsilon is equal to .000001, otherwise the default value is .0001. If lambda_search is set to True, the conditional values above are 1E-8 and 1E-6 respectively.",
level=expert)
public double gradient_epsilon
@API(help="Likelihood divider in objective value computation, default (of -1.0) will set it to 1/nobs.") public double obj_reg
@API(help="Link function.",
level=secondary,
values={"family_default","identity","logit","log","inverse","tweedie","ologit"})
public GLMModel.GLMParameters.Link link
@API(help="Method used to estimate the dispersion parameter for Tweedie, Gamma and Negative Binomial only.",
level=secondary,
values={"deviance","pearson","ml"})
public GLMModel.GLMParameters.DispersionMethod dispersion_parameter_method
@API(help="double array to initialize coefficients for GLM. If standardize is true, the standardized coefficients should be used. Otherwise, use the regular coefficients.",
gridable=true)
public double[] startval
@API(help="if true, will return likelihood function value.") public boolean calc_like
@API(help="if true, will generate variable inflation factors for numerical predictors. Default to false.",
level=expert)
public boolean generate_variable_inflation_factors
@API(help="Include constant term in the model",
level=expert)
public boolean intercept
@API(help="If set, will build a model with only the intercept. Default to false.",
level=expert)
public boolean build_null_model
@API(help="Only used for Tweedie, Gamma and Negative Binomial GLM. If set, will use the dispsersion parameter in init_dispersion_parameter as the standard error and use it to calculate the p-values. Default to false.",
level=expert)
public boolean fix_dispersion_parameter
@API(help="Only used for Tweedie, Gamma and Negative Binomial GLM. Store the initial value of dispersion parameter. If fix_dispersion_parameter is set, this value will be used in the calculation of p-values.",
level=expert,
gridable=true)
public double init_dispersion_parameter
@API(help="Prior probability for y==1. To be used only for logistic regression iff the data has been sampled and the mean of response does not reflect reality.",
level=expert)
public double prior
@API(help="Minimum lambda used in lambda search, specified as a ratio of lambda_max (the smallest lambda that drives all coefficients to zero). Default indicates: if the number of observations is greater than the number of variables, then lambda_min_ratio is set to 0.0001; if the number of observations is less than the number of variables, then lambda_min_ratio is set to 0.01.",
level=expert)
public double lambda_min_ratio
@API(help="Beta constraints",
direction=INPUT)
public water.api.schemas3.KeyV3.FrameKeyV3 beta_constraints
@API(help="Linear constraints: used to specify linear constraints involving more than one coefficients in standard form. It is only supported for solver IRLSM. It contains four columns: names (strings for coefficient names or constant), values, types ( strings of \'Equal\' or \'LessThanEqual\'), constraint_numbers (0 for first linear constraint, 1 for second linear constraint, ...).",
direction=INPUT)
public water.api.schemas3.KeyV3.FrameKeyV3 linear_constraints
@API(help="Maximum number of active predictors during computation. Use as a stopping criterion to prevent expensive model building with many predictors. Default indicates: If the IRLSM solver is used, the value of max_active_predictors is set to 5000 otherwise it is set to 100000000.",
direction=INPUT,
level=expert)
public int max_active_predictors
@API(help="A list of predictor column indices to interact. All pairwise combinations will be computed for the list.",
direction=INPUT,
level=expert)
public java.lang.String[] interactions
@API(help="A list of pairwise (first order) column interactions.",
direction=INPUT,
level=expert)
public water.api.schemas3.StringPairV3[] interaction_pairs
@API(help="Balance training data class counts via over/under-sampling (for imbalanced data).",
level=secondary,
direction=INOUT)
public boolean balance_classes
@API(help="Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes.",
level=expert,
direction=INOUT)
public float[] class_sampling_factors
@API(help="Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes.",
level=expert,
direction=INOUT)
public float max_after_balance_size
@API(help="[Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs.",
level=secondary,
direction=INOUT)
public int max_confusion_matrix_size
@API(help="Request p-values computation, p-values work only with IRLSM solver.",
level=secondary)
public boolean compute_p_values
@API(help="If true, will fix tweedie variance power value to the value set in tweedie_variance_power.",
level=secondary,
direction=INPUT)
public boolean fix_tweedie_variance_power
@API(help="In case of linearly dependent columns, remove the dependent columns.",
level=secondary)
public boolean remove_collinear_columns
@API(help="If changes in dispersion parameter estimation or loglikelihood value is smaller than dispersion_epsilon, will break out of the dispersion parameter estimation loop using maximum likelihood.",
level=secondary,
direction=INOUT)
public double dispersion_epsilon
@API(help="In estimating tweedie dispersion parameter using maximum likelihood, this is used to choose the lower and upper indices in the approximating of the infinite series summation.",
level=secondary,
direction=INOUT)
public double tweedie_epsilon
@API(help="Control the maximum number of iterations in the dispersion parameter estimation loop using maximum likelihood.",
level=secondary,
direction=INOUT)
public int max_iterations_dispersion
@API(help="If set to true, will generate scoring history for GLM. This may significantly slow down the algo.",
level=secondary,
direction=INPUT)
public boolean generate_scoring_history
@API(help="If true, will initialize coefficients with values derived from GLM runs without linear constraints. Only available for linear constraints.",
level=secondary,
direction=INOUT,
gridable=true)
public boolean init_optimal_glm
@API(help="If true, will keep the beta constraints and linear constraints separate. After new coefficients are found, first beta constraints will be applied followed by the application of linear constraints. Note that the beta constraints in this case will not be part of the objective function. If false, will combine the beta and linear constraints.",
level=secondary,
direction=INOUT,
gridable=true)
public boolean separate_linear_beta
@API(help="For constrained GLM only. It affects the setting of eta_k+1=eta_0/power(ck+1, alpha).",
level=expert,
direction=INOUT,
gridable=true)
public double constraint_eta0
@API(help="For constrained GLM only. It affects the setting of c_k+1=tau*c_k.",
level=expert,
direction=INOUT,
gridable=true)
public double constraint_tau
@API(help="For constrained GLM only. It affects the setting of eta_k = eta_0/pow(c_0, alpha).",
level=expert,
direction=INOUT,
gridable=true)
public double constraint_alpha
@API(help="For constrained GLM only. It affects the setting of eta_k+1 = eta_k/pow(c_k, beta).",
level=expert,
direction=INOUT,
gridable=true)
public double constraint_beta
@API(help="For constrained GLM only. It affects the initial setting of epsilon_k = 1/c_0.",
level=expert,
direction=INOUT,
gridable=true)
public double constraint_c0