Serialized Form

Package smile.anomaly

Class smile.anomaly.IsolationForest

class IsolationForest extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

c

double c

The normalizing factor.

extensionLevel

int extensionLevel

The extension level, i.e. how many dimension are specified in the random slope. With 0 extension level, it is coincident with the standard Isolation Forest.

trees

IsolationTree[] trees

Forest of isolation trees.

Class smile.anomaly.IsolationTree

class IsolationTree extends Object implements Serializable

Serialized Fields

root

smile.anomaly.IsolationTree.Node root

Tree root node.

Class smile.anomaly.SVM

class SVM extends KernelMachine<T> implements Serializable

Package smile.base.cart

Class smile.base.cart.CART

class CART extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

formula

smile.data.formula.Formula formula

The model formula.

importance

double[] importance

Variable importance. Every time a split of a node is made on variable the (GINI, information gain, etc.) impurity criterion for the two descendent nodes is less than the parent node. Adding up the decreases for each individual variable over the tree gives a simple measure of variable importance.

maxDepth

int maxDepth

The maximum depth of the tree.

maxNodes

int maxNodes

The maximum number of leaf nodes in the tree.

mtry

int mtry

The number of input variables to be used to determine the decision at a node of the tree.

nodeSize

int nodeSize

The number of instances in a node below which the tree will not split, setting nodeSize = 5 generally gives good results.

response

smile.data.type.StructField response

The schema of response variable.

root

Node root

The root of decision tree.

schema

smile.data.type.StructType schema

The schema of predictors.

Class smile.base.cart.DecisionNode

class DecisionNode extends LeafNode implements Serializable

serialVersionUID:

2L

Serialized Fields

count

int[] count

The number of node samples in each class.

output

int output

The predicted output.

Class smile.base.cart.InternalNode

class InternalNode extends Object implements Serializable

Serialized Fields

deviance

double deviance

The deviance of node.

falseChild

Node falseChild

Children node.

feature

int feature

The split feature for this node.

score

double score

Reduction in impurity compared to parent.

size

int size

The number of samples in the node.

trueChild

Node trueChild

Children node.

Class smile.base.cart.LeafNode

class LeafNode extends Object implements Serializable

Serialized Fields

size

int size

The number of samples in the node.

Class smile.base.cart.NominalNode

class NominalNode extends InternalNode implements Serializable

serialVersionUID:

2L

Serialized Fields

value

int value

The split value.

Class smile.base.cart.OrdinalNode

class OrdinalNode extends InternalNode implements Serializable

serialVersionUID:

2L

Serialized Fields

value

double value

The split value.

Class smile.base.cart.RegressionNode

class RegressionNode extends LeafNode implements Serializable

serialVersionUID:

2L

Serialized Fields

mean

double mean

The mean of response variable.

output

double output

The predicted output. In standard regression tree, this is same as the mean. However, in gradient tree boosting, this may be different.

rss

double rss

The residual sum of squares.

Package smile.base.mlp

Class smile.base.mlp.HiddenLayer

class HiddenLayer extends Layer implements Serializable

serialVersionUID:

2L

Serialized Fields

activation

ActivationFunction activation

The activation function.

Class smile.base.mlp.InputLayer

class InputLayer extends Layer implements Serializable

serialVersionUID:

2L

Serialization Methods

readObject

private void readObject(ObjectInputStream in)
                 throws IOException,
ClassNotFoundException

Initializes the workspace when deserializing the object.

Parameters:

in - the input stream.

Throws:

IOException - when fails to read the stream.

ClassNotFoundException - when fails to load the class.

Class smile.base.mlp.Layer

class Layer extends Object implements Serializable

serialVersionUID:

2L

Serialization Methods

readObject

private void readObject(ObjectInputStream in)
                 throws IOException,
ClassNotFoundException

Initializes the workspace when deserializing the object.

Parameters:

in - the input stream.

Throws:

IOException - when fails to read the stream.

ClassNotFoundException - when fails to load the class.

Serialized Fields

bias

double[] bias

The bias.

dropout

double dropout

The dropout rate. Dropout randomly sets input units to 0 with this rate at each step during training time, which helps prevent overfitting.

n

int n

The number of neurons in this layer

p

int p

The number of input variables.

weight

smile.math.matrix.Matrix weight

The affine transformation matrix.

Class smile.base.mlp.MultilayerPerceptron

class MultilayerPerceptron extends Object implements Serializable

serialVersionUID:

2L

Serialization Methods

readObject

private void readObject(ObjectInputStream in)
                 throws IOException,
ClassNotFoundException

Initializes the workspace when deserializing the object.

Parameters:

in - the input stream.

Throws:

IOException - when fails to read the stream.

ClassNotFoundException - when fails to load the class.

Serialized Fields

clipNorm

double clipNorm

The gradient clipping norm.

clipValue

double clipValue

The gradient clipping value.

epsilon

double epsilon

A small constant for numerical stability in RMSProp.

lambda

double lambda

The L2 regularization factor, which is also the weight decay factor.

learningRate

smile.math.TimeFunction learningRate

The learning rate.

momentum

smile.math.TimeFunction momentum

The momentum factor.

net

Layer[] net

The input and hidden layers.

output

OutputLayer output

The output layer.

p

int p

The dimensionality of input data.

rho

double rho

The discounting factor for the history/coming gradient in RMSProp.

t

int t

The training iterations.

Class smile.base.mlp.OutputLayer

class OutputLayer extends Layer implements Serializable

serialVersionUID:

2L

Serialized Fields

activation

OutputFunction activation

The output activation function.

cost

Cost cost

The cost function.

Package smile.base.rbf

Class smile.base.rbf.RBF

class RBF extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

center

T center

The center of neuron.

distance

smile.math.distance.Metric<T> distance

Metric distance.

rbf

smile.math.rbf.RadialBasisFunction rbf

Radial basis function.

Package smile.base.svm

Class smile.base.svm.KernelMachine

class KernelMachine extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

b

double b

The intercept.

kernel

smile.math.kernel.MercerKernel<T> kernel

The kernel function.

vectors

T[] vectors

The support vectors (or control points).

w

double[] w

The linear weights.

Class smile.base.svm.LASVM

class LASVM extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

b

double b

Threshold of decision function.

Cn

double Cn

The soft margin penalty parameter for negative samples.

Cp

double Cp

The soft margin penalty parameter for positive samples.

gmax

double gmax

The gradient of most violating pair.

gmin

double gmin

The gradient of most violating pair.

K

double[][] K

The kernel matrix.

kernel

smile.math.kernel.MercerKernel<T> kernel

The kernel function.

minmaxflag

boolean minmaxflag

True if minmax() is already called after update.

svmax

SupportVector<T> svmax

The most violating pair.

svmin

SupportVector<T> svmin

The most violating pair.

tol

double tol

The tolerance of convergence test.

vectors

ArrayList<SupportVector<T>> vectors

Support vectors.

x

T[] x

The training samples.

Class smile.base.svm.LinearKernelMachine

class LinearKernelMachine extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

b

double b

The intercept.

w

double[] w

The weight vector.

Class smile.base.svm.SupportVector

class SupportVector extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

alpha

double alpha

Lagrangian multiplier of support vector.

cmax

double cmax

Upper bound of alpha.

cmin

double cmin

Lower bound of alpha.

g

double g

Gradient y - Kα.

i

int i

The index of support vector in training samples.

k

double k

Kernel value k(x, x)

x

T x

Support vector.

Package smile.classification

Class smile.classification.AbstractClassifier

class AbstractClassifier extends Object implements Serializable

Serialized Fields

classes

smile.util.IntSet classes

The class labels.

Class smile.classification.AdaBoost

class AdaBoost extends AbstractClassifier<smile.data.Tuple> implements Serializable

serialVersionUID:

2L

Serialized Fields

alpha

double[] alpha

The weight of each decision tree.

error

double[] error

The weighted error of each decision tree during training.

formula

smile.data.formula.Formula formula

The model formula.

importance

double[] importance

Variable importance. Every time a split of a node is made on variable the (GINI, information gain, etc.) impurity criterion for the two descendent nodes is less than the parent node. Adding up the decreases for each individual variable over all trees in the forest gives a fast variable importance that is often very consistent with the permutation importance measure.

k

int k

The number of classes.

trees

DecisionTree[] trees

Forest of decision trees.

Class smile.classification.ClassLabels

class ClassLabels extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

classes

smile.util.IntSet classes

The class labels.

k

int k

The number of classes.

ni

int[] ni

The number of samples per classes.

priori

double[] priori

The estimated priori probabilities.

y

int[] y

The sample class id in [0, k).

Class smile.classification.DecisionTree

class DecisionTree extends CART implements Serializable

serialVersionUID:

2L

Serialized Fields

classes

smile.util.IntSet classes

The class labels.

k

int k

The number of classes.

rule

SplitRule rule

The splitting rule.

Class smile.classification.DiscreteNaiveBayes

class DiscreteNaiveBayes extends AbstractClassifier<int[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

fixedPriori

boolean fixedPriori

If true, don't update the priori during learning.

k

int k

The number of classes.

logcondprob

double[][] logcondprob

The log conditional probabilities for document classification.

model

DiscreteNaiveBayes.Model model

The generation model of naive Bayes.

n

int n

The total number of documents.

nc

int[] nc

The number of documents in each class.

nt

int[] nt

The number of terms per class.

ntc

int[][] ntc

The number of each term per class.

p

int p

The number of independent variables.

priori

double[] priori

The priori probability of each class.

sigma

double sigma

Amount of add-k smoothing of evidence. By default, we use add-one or Laplace smoothing, which simply adds one to each count to eliminate zeros. Add-one smoothing can be interpreted as a uniform prior (each term occurs once for each class) that is then updated as evidence from the training data comes in.

Class smile.classification.FLD

class FLD extends AbstractClassifier<double[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

k

int k

The number of classes.

mean

double[] mean

Projected mean vector.

mu

double[][] mu

Projected class mean vectors.

p

int p

The dimensionality of data.

scaling

smile.math.matrix.Matrix scaling

Project matrix.

Class smile.classification.GradientTreeBoost

class GradientTreeBoost extends AbstractClassifier<smile.data.Tuple> implements Serializable

serialVersionUID:

2L

Serialized Fields

b

double b

The intercept for binary classification.

forest

RegressionTree[][] forest

Forest of regression trees for multi-class classification.

formula

smile.data.formula.Formula formula

The model formula.

importance

double[] importance

Variable importance. Every time a split of a node is made on variable the impurity criterion for the two descendent nodes is less than the parent node. Adding up the decreases for each individual variable over all trees in the forest gives a simple variable importance.

k

int k

The number of classes.

shrinkage

double shrinkage

The shrinkage parameter in (0, 1] controls the learning rate of procedure.

trees

RegressionTree[] trees

Forest of regression trees for binary classification.

Class smile.classification.IsotonicRegressionScaling

class IsotonicRegressionScaling extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

buckets

double[] buckets

The step-wise buckets of function values in ascending order.

prob

double[] prob

The probability of instances falling into the corresponding buckets.

Class smile.classification.KNN

class KNN extends AbstractClassifier<T> implements Serializable

serialVersionUID:

2L

Serialized Fields

k

int k

The number of neighbors for decision.

knn

smile.neighbor.KNNSearch<T,T> knn

The data structure for nearest neighbor search.

y

int[] y

The labels of training samples.

Class smile.classification.LDA

class LDA extends AbstractClassifier<double[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

eigen

double[] eigen

The eigen values of common variance matrix.

k

int k

The number of classes.

logppriori

double[] logppriori

The constant term of discriminant function of each class.

mu

double[][] mu

THe mean vectors of each class.

p

int p

The dimensionality of data.

priori

double[] priori

The a priori probabilities of each class.

scaling

smile.math.matrix.Matrix scaling

The eigen vectors of common covariance matrix, which transforms observations to discriminant functions, normalized so that common covariance matrix is spherical.

Class smile.classification.LogisticRegression

class LogisticRegression extends AbstractClassifier<double[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

eta

double eta

learning rate for stochastic gradient descent.

k

int k

The number of classes.

L

double L

The log-likelihood of learned model.

lambda

double lambda

Regularization factor.

p

int p

The dimension of input space.

Class smile.classification.LogisticRegression.Binomial

class Binomial extends LogisticRegression implements Serializable

Serialized Fields

w

double[] w

The linear weights.

Class smile.classification.LogisticRegression.Multinomial

class Multinomial extends LogisticRegression implements Serializable

Serialized Fields

w

double[][] w

The linear weights.

Class smile.classification.Maxent

class Maxent extends AbstractClassifier<int[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

eta

double eta

learning rate for stochastic gradient descent.

k

int k

The number of classes.

L

double L

The log-likelihood of learned model.

lambda

double lambda

Regularization factor.

p

int p

The dimension of input space.

Class smile.classification.Maxent.Binomial

class Binomial extends Maxent implements Serializable

Serialized Fields

w

double[] w

The linear weights.

Class smile.classification.Maxent.Multinomial

class Multinomial extends Maxent implements Serializable

Serialized Fields

w

double[][] w

The linear weights.

Class smile.classification.MLP

class MLP extends MultilayerPerceptron implements Serializable

serialVersionUID:

2L

Serialized Fields

classes

smile.util.IntSet classes

The class label encoder.

k

int k

The number of classes.

Class smile.classification.NaiveBayes

class NaiveBayes extends AbstractClassifier<double[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

k

int k

The number of classes.

p

int p

The number of independent variables.

priori

double[] priori

The priori probability of each class.

prob

smile.stat.distribution.Distribution[][] prob

The conditional distribution for general purpose naive Bayes classifier.

Class smile.classification.OneVersusOne

class OneVersusOne extends AbstractClassifier<T> implements Serializable

serialVersionUID:

2L

Serialized Fields

classifiers

Classifier<T>[][] classifiers

The binary classifier.

k

int k

The number of classes.

platt

PlattScaling[][] platt

The binary classifier.

Class smile.classification.OneVersusRest

class OneVersusRest extends AbstractClassifier<T> implements Serializable

serialVersionUID:

2L

Serialized Fields

classifiers

Classifier<T>[] classifiers

The binary classifier.

k

int k

The number of classes.

platt

PlattScaling[] platt

The probability estimation by Platt scaling.

Class smile.classification.PlattScaling

class PlattScaling extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

alpha

double alpha

The scaling parameter.

beta

double beta

The scaling parameter.

Class smile.classification.QDA

class QDA extends AbstractClassifier<double[]> implements Serializable

serialVersionUID:

2L

Serialized Fields

eigen

double[][] eigen

Eigen values of each covariance matrix.

k

int k

The number of classes.

logppriori

double[] logppriori

Constant term of discriminant function of each class.

mu

double[][] mu

Mean vectors of each class.

p

int p

The dimensionality of data.

priori

double[] priori

A priori probabilities of each class.

scaling

smile.math.matrix.Matrix[] scaling

Eigen vectors of each covariance matrix, which transforms observations to discriminant functions, normalized so that within groups covariance matrix is spherical.

Class smile.classification.RandomForest

class RandomForest extends AbstractClassifier<smile.data.Tuple> implements Serializable

serialVersionUID:

2L

Serialized Fields

formula

smile.data.formula.Formula formula

The model formula.

importance

double[] importance

Variable importance. Every time a split of a node is made on variable the (GINI, information gain, etc.) impurity criterion for the two descendent nodes is less than the parent node. Adding up the decreases for each individual variable over all trees in the forest gives a fast variable importance that is often very consistent with the permutation importance measure.

k

int k

The number of classes.

metrics

ClassificationMetrics metrics

The overall out-of-bag metrics, which are quite accurate given that enough trees have been grown (otherwise the OOB error estimate can bias upward).

models

RandomForest.Model[] models

Forest of decision trees. The second value is the accuracy of tree on the OOB samples, which can be used a weight when aggregating tree votes.

Class smile.classification.RandomForest.Model

class Model extends Object implements Serializable

Serialized Fields

metrics

ClassificationMetrics metrics

The performance metrics on out-of-bag samples.

tree

DecisionTree tree

The decision tree.

weight

double weight

The weight of tree, which can be used when aggregating tree votes.

Class smile.classification.RBFNetwork

class RBFNetwork extends AbstractClassifier<T> implements Serializable

serialVersionUID:

2L

Serialized Fields

k

int k

The number of classes.

normalized

boolean normalized

True to fit a normalized RBF network.

rbf

RBF<T>[] rbf

The radial basis function.

w

smile.math.matrix.Matrix w

The linear weights.

Class smile.classification.RDA

class RDA extends QDA implements Serializable

serialVersionUID:

2L

Class smile.classification.SparseLogisticRegression

class SparseLogisticRegression extends AbstractClassifier<smile.util.SparseArray> implements Serializable

serialVersionUID:

2L

Serialized Fields

eta

double eta

learning rate for stochastic gradient descent.

k

int k

The number of classes.

L

double L

The log-likelihood of learned model.

lambda

double lambda

Regularization factor.

p

int p

The dimension of input space.

Class smile.classification.SparseLogisticRegression.Binomial

class Binomial extends SparseLogisticRegression implements Serializable

Serialized Fields

w

double[] w

The linear weights.

Class smile.classification.SparseLogisticRegression.Multinomial

class Multinomial extends SparseLogisticRegression implements Serializable

Serialized Fields

w

double[][] w

The linear weights.

Class smile.classification.SVM

class SVM extends KernelMachine<T> implements Serializable

Package smile.clustering

Class smile.clustering.CentroidClustering

class CentroidClustering extends PartitionClustering implements Serializable

serialVersionUID:

2L

Serialized Fields

centroids

T[] centroids

The centroids of each cluster.

distortion

double distortion

The total distortion.

Class smile.clustering.CLARANS

class CLARANS extends CentroidClustering<T,T> implements Serializable

serialVersionUID:

2L

Serialized Fields

distance

smile.math.distance.Distance<T> distance

The lambda of distance measure.

Class smile.clustering.DBSCAN

class DBSCAN extends PartitionClustering implements Serializable

serialVersionUID:

2L

Serialized Fields

core

boolean[] core

The flag if the point is a core point (at least minPts points are within radius).

minPts

double minPts

The minimum number of points required to form a cluster

nns

smile.neighbor.RNNSearch<T,T> nns

Data structure for neighborhood search.

radius

double radius

The neighborhood radius.

Class smile.clustering.DENCLUE

class DENCLUE extends PartitionClustering implements Serializable

serialVersionUID:

2L

Serialized Fields

attractors

double[][] attractors

The density attractor of each observation.

radius

double[] radius

The radius of density attractor.

samples

double[][] samples

The samples decided by k-means used in the iterations of hill climbing.

sigma

double sigma

The smooth parameter in the Gaussian kernel.

tol

double tol

The tolerance of hill-climbing procedure.

Class smile.clustering.DeterministicAnnealing

class DeterministicAnnealing extends CentroidClustering<double[],double[]> implements Serializable

serialVersionUID:

2L

Class smile.clustering.GMeans

class GMeans extends CentroidClustering<double[],double[]> implements Serializable

serialVersionUID:

2L

Class smile.clustering.HierarchicalClustering

class HierarchicalClustering extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

height

double[] height

A set of n-1 non-decreasing real values, which are the clustering height, i.e., the value of the criterion associated with the clustering method for the particular agglomeration.

merge

int[][] merge

An n-1 by 2 matrix of which row i describes the merging of clusters at step i of the clustering. If an element j in the row is less than n, then observation j was merged at this stage. If j >= n then the merge was with the cluster formed at the (earlier) stage j-n of the algorithm.

Class smile.clustering.KMeans

class KMeans extends CentroidClustering<double[],double[]> implements Serializable

serialVersionUID:

2L

Class smile.clustering.KModes

class KModes extends CentroidClustering<int[],int[]> implements Serializable

serialVersionUID:

2L

Class smile.clustering.MEC

class MEC extends PartitionClustering implements Serializable

serialVersionUID:

2L

Serialized Fields

entropy

double entropy

The conditional entropy as the objective function.

nns

smile.neighbor.RNNSearch<T,T> nns

The neighborhood search data structure.

radius

double radius

The range of neighborhood.

Class smile.clustering.PartitionClustering

class PartitionClustering extends Object implements Serializable

Serialized Fields

k

int k

The number of clusters.

size

int[] size

The number of observations in each cluster.

y

int[] y

The cluster labels of data.

Class smile.clustering.SIB

class SIB extends CentroidClustering<double[],smile.util.SparseArray> implements Serializable

serialVersionUID:

2L

Class smile.clustering.SpectralClustering

class SpectralClustering extends PartitionClustering implements Serializable

serialVersionUID:

2L

Serialized Fields

distortion

double distortion

The distortion in feature space.

Class smile.clustering.XMeans

class XMeans extends CentroidClustering<double[],double[]> implements Serializable

serialVersionUID:

2L

Package smile.deep.activation

Class smile.deep.activation.LeakyReLU

class LeakyReLU extends Object implements Serializable

Serialized Fields

a

double a

The leaky parameter 0 <= a < 1.

Class smile.deep.activation.ReLU

class ReLU extends Object implements Serializable

Class smile.deep.activation.Sigmoid

class Sigmoid extends Object implements Serializable

Class smile.deep.activation.Softmax

class Softmax extends Object implements Serializable

Class smile.deep.activation.Tanh

class Tanh extends Object implements Serializable

Package smile.deep.optimizer

Class smile.deep.optimizer.Adam

class Adam extends Object implements Serializable

Serialized Fields

beta1

double beta1

The exponential decay rate for the 1st moment estimates.

beta2

double beta2

The exponential decay rate for the 2nd moment estimates.

epsilon

double epsilon

A small constant for numerical stability.

learningRate

smile.math.TimeFunction learningRate

The learning rate.

Class smile.deep.optimizer.RMSProp

class RMSProp extends Object implements Serializable

Serialized Fields

epsilon

double epsilon

A small constant for numerical stability.

learningRate

smile.math.TimeFunction learningRate

The learning rate.

rho

double rho

The discounting factor for the history/coming gradient.

Class smile.deep.optimizer.SGD

class SGD extends Object implements Serializable

Serialized Fields

learningRate

smile.math.TimeFunction learningRate

The learning rate.

momentum

smile.math.TimeFunction momentum

The momentum factor.

Package smile.feature.extraction

Class smile.feature.extraction.BagOfWords

class BagOfWords extends Object implements Serializable

Serialized Fields

binary

boolean binary

True to check if feature words appear in a document instead of their frequencies.

columns

String[] columns

The input text fields.

featureIndex

Map<String,Integer> featureIndex

The mapping from feature words to indices.

schema

smile.data.type.StructType schema

The schema of output space.

tokenizer

Function<String,String[]> tokenizer

The tokenizer of text, which may include additional processing such as filtering stop word, converting to lowercase, stemming, etc.

words

String[] words

The feature words.

Class smile.feature.extraction.GHA

class GHA extends Projection implements Serializable

serialVersionUID:

2L

Serialized Fields

n

int n

The dimension of input space.

p

int p

The dimension of feature space.

r

smile.math.TimeFunction r

The learning rate;

t

int t

The training iterations.

wy

double[] wy

Workspace for W' * y.

y

double[] y

Workspace for W * x.

Class smile.feature.extraction.KernelPCA

class KernelPCA extends Projection implements Serializable

serialVersionUID:

2L

Serialized Fields

kpca

KPCA<double[]> kpca

Kernel PCA.

Class smile.feature.extraction.PCA

class PCA extends Projection implements Serializable

serialVersionUID:

2L

Serialized Fields

cumulativeProportion

double[] cumulativeProportion

The cumulative proportion of variance contained in principal components.

eigvalues

double[] eigvalues

Eigenvalues of principal components.

eigvectors

smile.math.matrix.Matrix eigvectors

The matrix of variable loadings, whose columns contain the eigenvectors.

mu

double[] mu

The sample mean.

pmu

double[] pmu

The projected sample mean.

proportion

double[] proportion

The proportion of variance contained in each principal component.

Class smile.feature.extraction.ProbabilisticPCA

class ProbabilisticPCA extends Projection implements Serializable

serialVersionUID:

2L

Serialized Fields

loading

smile.math.matrix.Matrix loading

The loading matrix.

mu

double[] mu

The sample mean.

noise

double noise

The variance of noise part.

pmu

double[] pmu

The projected sample mean.

Class smile.feature.extraction.Projection

class Projection extends Object implements Serializable

Serialized Fields

columns

String[] columns

The fields of input space.

projection

smile.math.matrix.Matrix projection

The projection matrix. The dimension reduced data can be obtained by y = W * x.

schema

smile.data.type.StructType schema

The schema of output space.

Class smile.feature.extraction.RandomProjection

class RandomProjection extends Projection implements Serializable

serialVersionUID:

2L

Package smile.feature.imputation

Class smile.feature.imputation.KMedoidsImputer

class KMedoidsImputer extends Object implements Serializable

Serialized Fields

kmedoids

CLARANS<smile.data.Tuple> kmedoids

The K-Medoids clustering.

Class smile.feature.imputation.KNNImputer

class KNNImputer extends Object implements Serializable

Serialized Fields

k

int k

The number of nearest neighbors used for imputation.

knn

smile.neighbor.KNNSearch<smile.data.Tuple,smile.data.Tuple> knn

K-nearest neighbor search algorithm.

Class smile.feature.imputation.SimpleImputer

class SimpleImputer extends Object implements Serializable

Serialized Fields

values

Map<String,Object> values

The map of column name to the constant value.

Package smile.feature.transform

Class smile.feature.transform.Normalizer

class Normalizer extends Object implements Serializable

Serialized Fields

columns

Set<String> columns

The columns to transform.

norm

Normalizer.Norm norm

The vector norm.

Package smile.glm

Class smile.glm.GLM

class GLM extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

beta

double[] beta

The linear weights.

deviance

double deviance

The deviance = 2 * (LogLikelihood(Saturated Model) - LogLikelihood(Proposed Model)).

devianceResiduals

double[] devianceResiduals

The deviance residuals.

df

int df

The degrees of freedom of the residual deviance.

formula

smile.data.formula.Formula formula

The symbolic description of the model to be fitted.

logLikelihood

double logLikelihood

Log-likelihood.

model

Model model

The model specifications (link function, deviance, etc.).

mu

double[] mu

The fitted mean values.

nullDeviance

double nullDeviance

The null deviance = 2 * (LogLikelihood(Saturated Model) - LogLikelihood(Null Model)).

The saturated model, also referred to as the full model or maximal model, allows a different mean response for each group of replicates. One can think of the saturated model as having the most general possible mean structure for the data since the means are unconstrained.

The null model assumes that all observations have the same distribution with common parameter. Like the saturated model, the null model does not depend on predictor variables. While the saturated most is the most general model, the null model is the most restricted model.

predictors

String[] predictors

The predictors of design matrix.

ztest

double[][] ztest

The coefficients, their standard errors, z-scores, and p-values.

Package smile.glm.model

Package smile.manifold

Class smile.manifold.IsoMap

class IsoMap extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

coordinates

double[][] coordinates

The coordinate matrix in embedding space.

graph

smile.graph.AdjacencyList graph

The nearest neighbor graph.

index

int[] index

The original sample index.

Class smile.manifold.KPCA

class KPCA extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

coordinates

double[][] coordinates

The coordinates of projected training data.

data

T[] data

Training data.

kernel

smile.math.kernel.MercerKernel<T> kernel

Mercer kernel.

latent

double[] latent

The eigenvalues of kernel principal components.

mean

double[] mean

The row mean of kernel matrix.

mu

double mu

The mean of kernel matrix.

projection

smile.math.matrix.Matrix projection

The projection matrix.

Class smile.manifold.LaplacianEigenmap

class LaplacianEigenmap extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

coordinates

double[][] coordinates

The coordinate matrix in embedding space.

graph

smile.graph.AdjacencyList graph

Nearest neighbor graph.

index

int[] index

The original sample index.

width

double width

The width of heat kernel.

Class smile.manifold.LLE

class LLE extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

coordinates

double[][] coordinates

The coordinate matrix in embedding space.

graph

smile.graph.AdjacencyList graph

Nearest neighbor graph.

index

int[] index

The original sample index.

Class smile.manifold.TSNE

class TSNE extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

coordinates

double[][] coordinates

The coordinate matrix in embedding space.

cost

double cost

The cost function value.

eta

double eta

The learning rate.

finalMomentum

double finalMomentum

The momentum in later stage.

gains

double[][] gains

The gain matrix.

minGain

double minGain

The floor of gain.

momentum

double momentum

The momentum factor.

momentumSwitchIter

int momentumSwitchIter

The number of iterations at which switch the momentum to finalMomentum.

P

double[][] P

The probability matrix of the distances in the input space.

Q

double[][] Q

The probability matrix of the distances in the feature space.

Qsum

double Qsum

The sum of Q matrix.

totalIter

int totalIter

The number of iterations so far.

Class smile.manifold.UMAP

class UMAP extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

coordinates

double[][] coordinates

The coordinate matrix in embedding space.

graph

smile.graph.AdjacencyList graph

The nearest neighbor graph.

index

int[] index

The original sample index.

Package smile.regression

Class smile.regression.GaussianProcessRegression

class GaussianProcessRegression extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

cholesky

smile.math.matrix.Matrix.Cholesky cholesky

The Cholesky decomposition of kernel matrix.

kernel

smile.math.kernel.MercerKernel<T> kernel

The covariance/kernel function.

L

double L

The log marginal likelihood, which may be not available (NaN) when the model is fit with approximate methods.

mean

double mean

The mean of responsible variable.

noise

double noise

The variance of noise.

regressors

T[] regressors

The regressors.

sd

double sd

The standard deviation of responsible variable.

w

double[] w

The linear weights.

Class smile.regression.GradientTreeBoost

class GradientTreeBoost extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

b

double b

The intercept.

formula

smile.data.formula.Formula formula

The model formula.

importance

double[] importance

Variable importance. Every time a split of a node is made on variable the impurity criterion for the two descendent nodes is less than the parent node. Adding up the decreases for each individual variable over all trees in the forest gives a simple variable importance.

shrinkage

double shrinkage

The shrinkage parameter in (0, 1] controls the learning rate of procedure.

trees

RegressionTree[] trees

Forest of regression trees.

Class smile.regression.KernelMachine

class KernelMachine extends KernelMachine<T> implements Serializable

serialVersionUID:

2L

Class smile.regression.LinearModel

class LinearModel extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

adjustedRSquared

double adjustedRSquared

Adjusted R². The adjusted R² has almost same explanation as R² but it penalizes the statistic as extra variables are included in the model.

b

double b

The intercept.

bias

boolean bias

True if the linear weights w includes the intercept.

df

int df

The degree-of-freedom of residual standard error.

error

double error

Residual standard error.

F

double F

The F-statistic of the goodness-of-fit of the model.

fittedValues

double[] fittedValues

The fitted values.

formula

smile.data.formula.Formula formula

Design matrix formula

p

int p

The dimensionality.

predictors

String[] predictors

The predictors of design matrix.

pvalue

double pvalue

The p-value of the goodness-of-fit test of the model.

residuals

double[] residuals

The residuals, that is response minus fitted values.

RSquared

double RSquared

R². R² is a statistic that will give some information about the goodness of fit of a model. In regression, the R² coefficient of determination is a statistical measure of how well the regression line approximates the real data points. An R² of 1.0 indicates that the regression line perfectly fits the data.

In the case of ordinary least-squares regression, R² increases as we increase the number of variables in the model (R² will not decrease). This illustrates a drawback to one possible use of R², where one might try to include more variables in the model until "there is no more improvement". This leads to the alternative approach of looking at the adjusted R².

RSS

double RSS

Residual sum of squares.

schema

smile.data.type.StructType schema

The schema of design matrix.

ttest

double[][] ttest

The coefficients, their standard errors, t-scores, and p-values.

V

smile.math.matrix.Matrix V

First initialized to the matrix (X^TX)^-1, it is updated with each new learning instance.

w

double[] w

The linear weights.

Class smile.regression.MLP

class MLP extends MultilayerPerceptron implements Serializable

serialVersionUID:

2L

Serialized Fields

scaler

smile.math.Scaler scaler

The scaling function of output values.

Class smile.regression.RandomForest

class RandomForest extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

formula

smile.data.formula.Formula formula

The model formula.

importance

double[] importance

Variable importance. Every time a split of a node is made on variable the impurity criterion for the two descendent nodes is less than the parent node. Adding up the decreases for each individual variable over all trees in the forest gives a fast variable importance that is often very consistent with the permutation importance measure.

metrics

RegressionMetrics metrics

The overall out-of-bag metrics, which are quite accurate given that enough trees have been grown (otherwise the OOB error estimate can bias upward).

models

RandomForest.Model[] models

Forest of regression trees.

Class smile.regression.RandomForest.Model

class Model extends Object implements Serializable

Serialized Fields

metrics

RegressionMetrics metrics

The performance metrics on out-of-bag samples.

tree

RegressionTree tree

The decision tree.

Class smile.regression.RBFNetwork

class RBFNetwork extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

normalized

boolean normalized

True to fit a normalized RBF network.

rbf

RBF<T>[] rbf

The radial basis functions.

w

double[] w

The linear weights.

Class smile.regression.RegressionTree

class RegressionTree extends CART implements Serializable

serialVersionUID:

2L

Package smile.sequence

Class smile.sequence.CRF

class CRF extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

potentials

RegressionTree[][] potentials

The potential functions for each class.

schema

smile.data.type.StructType schema

The schema of (x, s_j).

shrinkage

double shrinkage

The learning rate.

Class smile.sequence.CRFLabeler

class CRFLabeler extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

features

Function<T,smile.data.Tuple> features

The feature function.

model

CRF model

The CRF model.

Class smile.sequence.HMM

class HMM extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

a

smile.math.matrix.Matrix a

State transition probabilities.

b

smile.math.matrix.Matrix b

Symbol emission probabilities.

pi

double[] pi

Initial state probabilities.

Class smile.sequence.HMMLabeler

class HMMLabeler extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

model

HMM model

The HMM model.

ordinal

ToIntFunction<T> ordinal

The lambda returns the ordinal numbers of symbols.

Package smile.timeseries

Class smile.timeseries.AR

class AR extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

adjustedR2

double adjustedR2

Adjusted R². The adjusted R² has almost same explanation as R² but it penalizes the statistic as extra variables are included in the model.

ar

double[] ar

The linear weights of AR.

b

double b

The intercept.

df

int df

The degree-of-freedom of residual variance.

fittedValues

double[] fittedValues

The fitted values.

mean

double mean

The mean of time series.

method

AR.Method method

The fitting method.

p

int p

The order.

R2

double R2

R². R² is a statistic that will give some information about the goodness of fit of a model. In regression, the R² coefficient of determination is a statistical measure of how well the regression line approximates the real data points. An R² of 1.0 indicates that the regression line perfectly fits the data.

In the case of ordinary least-squares regression, R² increases as we increase the number of variables in the model (R² will not decrease). This illustrates a drawback to one possible use of R², where one might try to include more variables in the model until "there is no more improvement". This leads to the alternative approach of looking at the adjusted R².

residuals

double[] residuals

The residuals, that is response minus fitted values.

RSS

double RSS

Residual sum of squares.

ttest

double[][] ttest

The coefficients, their standard errors, t-scores, and p-values.

variance

double variance

Estimated variance.

x

double[] x

The time series.

Class smile.timeseries.ARMA

class ARMA extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

adjustedR2

double adjustedR2

Adjusted R². The adjusted R² has almost same explanation as R² but it penalizes the statistic as extra variables are included in the model.

ar

double[] ar

The linear weights of AR.

b

double b

The intercept.

df

int df

The degree-of-freedom of residual variance.

fittedValues

double[] fittedValues

The fitted values.

ma

double[] ma

The linear weights of MA.

mean

double mean

The mean of time series.

p

int p

The order of AR.

q

int q

The order of MA.

R2

double R2

R². R² is a statistic that will give some information about the goodness of fit of a model. In regression, the R² coefficient of determination is a statistical measure of how well the regression line approximates the real data points. An R² of 1.0 indicates that the regression line perfectly fits the data.

In the case of ordinary least-squares regression, R² increases as we increase the number of variables in the model (R² will not decrease). This illustrates a drawback to one possible use of R², where one might try to include more variables in the model until "there is no more improvement". This leads to the alternative approach of looking at the adjusted R².

residuals

double[] residuals

The residuals, that is response minus fitted values.

RSS

double RSS

Residual sum of squares.

ttest

double[][] ttest

The coefficients, their standard errors, t-scores, and p-values.

variance

double variance

Estimated variance.

x

double[] x

The time series.

Package smile.validation

Class smile.validation.Bag

class Bag extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

oob

int[] oob

The index of testing instances.

samples

int[] samples

The random samples.

Class smile.validation.ClassificationMetrics

class ClassificationMetrics extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

accuracy

double accuracy

The accuracy on validation data.

auc

double auc

The AUC on validation data.

crossentropy

double crossentropy

The cross entropy on validation data.

error

int error

The number of errors.

f1

double f1

The F-1 score on validation data.

fitTime

double fitTime

The time in milliseconds of fitting the model.

logloss

double logloss

The log loss on validation data.

mcc

double mcc

The Matthews correlation coefficient on validation data.

precision

double precision

The precision on validation data.

scoreTime

double scoreTime

The time in milliseconds of scoring the validation data.

sensitivity

double sensitivity

The sensitivity on validation data.

size

int size

The validation data size.

specificity

double specificity

The specificity on validation data.

Class smile.validation.ClassificationValidation

class ClassificationValidation extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

confusion

ConfusionMatrix confusion

The confusion matrix.

metrics

ClassificationMetrics metrics

The classification metrics.

model

M model

The model.

posteriori

double[][] posteriori

The posteriori probability of prediction if the model is a soft classifier.

prediction

int[] prediction

The model prediction.

truth

int[] truth

The true class labels of validation data.

Class smile.validation.ClassificationValidations

class ClassificationValidations extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

avg

ClassificationMetrics avg

The average of metrics.

rounds

List<ClassificationValidation<M>> rounds

The multiple round validations.

sd

ClassificationMetrics sd

The standard deviation of metrics.

Class smile.validation.RegressionMetrics

class RegressionMetrics extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

fitTime

double fitTime

The time in milliseconds of fitting the model.

mad

double mad

The mean absolute deviation on validation data.

mse

double mse

The mean squared error on validation data.

r2

double r2

The R-squared score on validation data.

rmse

double rmse

The root mean squared error on validation data.

rss

double rss

The residual sum of squares on validation data.

scoreTime

double scoreTime

The time in milliseconds of scoring the validation data.

size

int size

The validation data size.

Class smile.validation.RegressionValidation

class RegressionValidation extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

metrics

RegressionMetrics metrics

The regression metrics.

model

M model

The model.

prediction

double[] prediction

The model prediction.

truth

double[] truth

The true response variable of validation data.

Class smile.validation.RegressionValidations

class RegressionValidations extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

avg

RegressionMetrics avg

The average of metrics.

rounds

List<RegressionValidation<M>> rounds

The multiple round validations.

sd

RegressionMetrics sd

The standard deviation of metrics.

Package smile.validation.metric

Class smile.validation.metric.Accuracy

class Accuracy extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.AdjustedMutualInformation

class AdjustedMutualInformation extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

method

AdjustedMutualInformation.Method method

The normalization method.

Class smile.validation.metric.AdjustedRandIndex

class AdjustedRandIndex extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.AUC

class AUC extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.ConfusionMatrix

class ConfusionMatrix extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

matrix

int[][] matrix

Confusion matrix.

Class smile.validation.metric.Error

class Error extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.Fallout

class Fallout extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.FDR

class FDR extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.FScore

class FScore extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

beta

double beta

A positive value such that F-score measures the effectiveness of retrieval with respect to a user who attaches β times as much importance to recall as precision. The default value 1.0 corresponds to F1-score.

Class smile.validation.metric.LogLoss

class LogLoss extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.MAD

class MAD extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.MatthewsCorrelation

class MatthewsCorrelation extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.MSE

class MSE extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.MutualInformation

class MutualInformation extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.NormalizedMutualInformation

class NormalizedMutualInformation extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

method

NormalizedMutualInformation.Method method

The normalization method.

Class smile.validation.metric.Precision

class Precision extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.R2

class R2 extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.RandIndex

class RandIndex extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.Recall

class Recall extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.RMSE

class RMSE extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.RSS

class RSS extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.Sensitivity

class Sensitivity extends Object implements Serializable

serialVersionUID:

2L

Class smile.validation.metric.Specificity

class Specificity extends Object implements Serializable

serialVersionUID:

2L

Package smile.vq

Class smile.vq.BIRCH

class BIRCH extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

B

int B

The branching factor of non-leaf nodes.

d

int d

The dimensionality of data.

L

int L

The number of CF entries in the leaf nodes.

root

smile.vq.BIRCH.Node root

The root of CF tree.

T

double T

THe maximum radius of a sub-cluster.

Class smile.vq.GrowingNeuralGas

class GrowingNeuralGas extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

alpha

double alpha

Decrease error variables by multiplying them with alpha during inserting a new neuron.

beta

double beta

Decrease all error variables by multiply them with beta.

d

int d

The dimensionality of signals.

edgeLifetime

int edgeLifetime

The maximum age of edges.

epsBest

double epsBest

The learning rate to update best matching neuron.

epsNeighbor

double epsNeighbor

The learning rate to update neighbors of best matching neuron.

lambda

int lambda

If the number of input signals so far is an integer multiple of lambda, insert a new neuron.

neurons

ArrayList<Neuron> neurons

Neurons in the neural network.

t

int t

The number of signals processed so far.

top2

Neuron[] top2

The workspace to find nearest neighbors.

Class smile.vq.NeuralGas

class NeuralGas extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

alpha

smile.math.TimeFunction alpha

The learning rate function.

dist

double[] dist

The distance between a new observation to neurons.

eps

double eps

The threshold to update neuron if alpha * theta > eps.

graph

smile.graph.AdjacencyMatrix graph

The network of neurons.

lifetime

smile.math.TimeFunction lifetime

The lifetime of connections.

neurons

smile.vq.NeuralGas.Neuron[] neurons

The neurons.

t

int t

The current iteration.

theta

smile.math.TimeFunction theta

The neighborhood function.

Class smile.vq.NeuralMap

class NeuralMap extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

beta

double beta

Decrease the freshness of all neurons by multiply them with beta.

edgeLifetime

int edgeLifetime

The maximum age of edges.

epsBest

double epsBest

The learning rate to update nearest neuron.

epsNeighbor

double epsNeighbor

The learning to update neighbors of nearest neuron.

neurons

ArrayList<Neuron> neurons

Neurons in the neural network.

r

double r

The distance radius to activate a neuron for a given signal.

t

int t

The number of signals processed so far.

top2

Neuron[] top2

The workspace to find nearest neighbors.

Class smile.vq.SOM

class SOM extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

alpha

smile.math.TimeFunction alpha

The learning rate function.

dist

double[] dist

The distance between a new observation to neurons.

map

smile.vq.SOM.Neuron[][] map

The lattice of neurons.

ncol

int ncol

The number of columns in the lattice.

neurons

smile.vq.SOM.Neuron[] neurons

The neurons in linear array.

nrow

int nrow

The number of rows in the lattice.

t

int t

The current iteration.

theta

Neighborhood theta

The neighborhood function.

tol

double tol

The threshold to update neuron if alpha * theta > eps.

Package smile.vq.hebb

Class smile.vq.hebb.Edge

class Edge extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

age

int age

The age of the edges.

neighbor

Neuron neighbor

The neighbor neuron.

Class smile.vq.hebb.Neuron

class Neuron extends Object implements Serializable

serialVersionUID:

2L

Serialized Fields

counter

double counter

The local counter variable (e.g. the accumulated error, freshness, etc.)

edges

List<Edge> edges

The direct connected neighbors.

w

double[] w

The reference vector.