Global

Methods

accuracy_score(y_true, y_pred)

In multilabel classification, this function computes subset accuracy: the set of labels predicted for a sample must *exactly* match the corresponding set of labels in y_true.
Parameters:
Name Type Description
y_true Array 1d array-like, or label indicator array; Ground truth (correct) labels
y_pred Array 1d array-like, or label indicator array; Predicted labels, as returned by a classifier.
Source:

check_1dy(y, make_tf)

Check if outputs y are of proper format.
Parameters:
Name Type Default Description
y Array Outputs.
make_tf Boolean true Whether to convert y to tf.tensor
Source:

check_2dX(X, make_tf)

Check if inputs X are of proper format.
Parameters:
Name Type Default Description
X Array Input samples.
make_tf Boolean true Whether to convert input samples to tf.tensor
Source:

check_2dXy(X, y, make_tf)

Check whether a dataset is consistent, and whether inputs are a matrix and outputs are a vector.
Parameters:
Name Type Default Description
X Array Array of input observations.
y Array Outputs.
make_tf Boolean true Whether to convert y to tf.tensor
Source:

check_array(array, min_nd, max_nd, min_shape, max_shape, make_tf)

Check dimensions of the array.
Parameters:
Name Type Default Description
array Array Array to check, and optionally convert to tf tensor.
min_nd Number 2 minimum number of the dimensions in the array.
max_nd Number 2 maximum number of dimensions.
min_shape Array array of maximal sizes for dimensions of the tensor.
max_shape Array null maximal sizes of dimensions in array.
make_tf Boolean true whether to convert the input array to tf tensor.
Source:

check_is_fitted(estimator, attributes, all_or_any)

Checks if the estimator is fitted by verifying the presence of “all_or_any” of the passed attributes and raises an Error with the given message.
Parameters:
Name Type Default Description
estimator Object estimator to check to be fitted.
attributes Array list of attributes to check presence of.
all_or_any String all Specify whether all or any of the given attributes must exist.
Source:

ClassifierMixin(superclass)

Mixin class for all classifiers.
Parameters:
Name Type Description
superclass class Class to extend with functionality
Source:

(async) clone(obj, stringify)

Creates a copy of the object by serializing the object to json, and then deserializing.
Parameters:
Name Type Default Description
obj Any object to be cloned.
stringify Boolean false whether to stringify obtained blueprint. Should only be used for testing of whether the serialization representation is convertable to string, or in case serialized object might contain pointers to nested objects of the parent object, which erroneously are not cloned properly.
Source:

(async) cross_val_score(estimator, X, y, groups, scoring, cv) → {Array}

Evaluate estimator performance via cross - validation. Interface similar to `cross_val_score` function of sklearn.
Parameters:
Name Type Default Description
estimator BaseEstimator Model to be scored.
X Array Input values to be used for sorting.
y Array null Output values to be used for scoring
groups Object null Ignored.
scoring Object null Ignored for the time being.
cv Number 3 Number of folds used for cross - validation.
Source:
Returns:
Scores on various partitions.
Type
Array

(async) dumpjson(obj)

Convert various objects to json format. Is useful for serialization and deserialization of aitable objects.
Parameters:
Name Type Description
obj Any Instance of an object to be converted into a serializable json. Can be a json serializable value, a class that can be serialized, or instance of tensorflowjs objects.
Source:

(async) loadjson(json)

Inverse of dumpjson.
Parameters:
Name Type Description
json Any Blueprint of the object, to be deserialized.
Source:

make_murmaid(root, features, definitions, links)

Converts a decision tree, given by the root of the tree, into a text format that can be rendered to the user.
Parameters:
Name Type Default Description
root Object Root node of a decision tree.
features Array null Array of dictionaries, which contain information on names of the features and type of the features.
definitions Array null Is only used for recurrent calls inside
links Array null Similarily, stores the links between elements in graph.
Source:

r2_score(y_true, y_pred, sample_weight)

R^2 (coefficient of determination) regression score function. Best possible score is 1.0 and it can be negative (because the model can be arbitrarily worse). A constant model that always predicts the expected value of y, disregarding the input features, would get a R^2 score of 0.0.
Parameters:
Name Type Description
y_true Array array-like of shape = (n_samples); Ground truth (correct) target values.
y_pred Array array-like of shape = (n_samples) or (n_samples, n_outputs); Estimated target values.
sample_weight Array array-like of shape = (n_samples), optional; Sample weights.
Source:

random_partitioning(N, weights) → {Array}

Generate index set for partitioning of array of size N into a number of partitions. The size of partitions can be specified with `partitions` argument. The function ensures that the minimal size of every partition is at least one. This is important for small sized data partitioning, for example, or in case of leave one out cross - validation.
Parameters:
Name Type Description
N Integer Size of array to be partitioned
weights Array Weights that are assigned to partitions. With higher weight, the partition is more likely to receive elements.
Source:
Returns:
Every element in the output array is a set of indicies of the elements that belong to the i-th partition.
Type
Array

RegressorMixin(superclass)

Mixin class for all regressors.
Parameters:
Name Type Description
superclass class Class to extend with functionality
Source:

(async) score(X, y) → {Number}

Returns the mean accuracy on the given test data and labels. In multi-label classification, this is the subset accuracy which is a harsh metric since you require for each sample that each label set be correctly predicted.
Parameters:
Name Type Description
X Array Test samples.
y Array Test labels for X.
Source:
Returns:
Mean accuracy of this.predict(X) wrt. y.
Type
Number

(async) score(X, y) → {Number}

Returns the coefficient of determination R^2 of the prediction. The coefficient R^2 is defined as (1 - u/v), where u is the residual sum of squares ((y_true - y_pred) ** 2).sum() and v is the total sum of squares ((y_true - y_true.mean()) ** 2).sum(). The best possible score is 1.0 and it can be negative (because the model can be arbitrarily worse). A constant model that always predicts the expected value of y, disregarding the input features, would get a R^2 score of 0.0.
Parameters:
Name Type Description
X Array Test samples.
y Array Test outputs for X.
Source:
Returns:
R^2 of this.predict(X) wrt. y.
Type
Number

set_defaults(params, defaults)

Sets default parameter values for the model classes.
Parameters:
Name Type Default Description
params Dictionary null Parameters supplied by user
defaults Dictionary null Default values of parameters
Source:

shuffle(a)

Shuffles array in place.
Parameters:
Name Type Description
a Array items An array containing the items.
Source:

t1d(data)

Converts some TypedArray into 1d tfjs array.
Parameters:
Name Type Description
data Array data to be converted to 1d tfjs array
Source:

t2d(data)

Converts some TypedArray into 2d tfjs array.
Parameters:
Name Type Description
data Array data to be converted to 2d tfjs array
Source:

train_test_split(arrays, test_split)

Splits the data into training and testing partitions for model fitting. The split is done at random.
Parameters:
Name Type Default Description
arrays Array List of arrays to be split into training and testing parts.
test_split Number 0.25 Value in the range of [0; 1]. Specifies how much of the data will be used for test partition.
Source: