The Abs operator.

AbsolutePearson measures the ROC AUC score.

Accuracy measures the fraction of matches between true classes and predicted classes.

The Add operator.

CodeDisplay outputs an code-like representation of an Operator.

A ConfusionMatrix stores true positives (TP), true negatives (TN), false positives (FP) and false negatives (FN).

The Const operator always returns the same value.

The Cos operator.

A Crossover takes two Operators and combines them in order to produce two new Operators.

A DiffMetric is a Metric that can compute element-wise gradients.

DirDisplay outputs a directory like representation of an Operator.

A Displayer outputs a string representation of a Operator.

The Div operator.

F1 measures the F1-score.

FullInit can generate an Operator of height maxHeight.

An GP links all the different components together and can be used to train Programs on a dataset. You shouldn't instantiate this struct directly; inst

A GPConfig contains all the information needed to instantiate an GP.

GoldenLineSearch implements Golden-section line-search.

GradientBoosting implements gradient boosting on top of genetic programming.

GraphvizDisplay outputs a Graphviz representation of an Operator. Each branch is indexed with a global counter and is labelled with the Operator's Nam

GrowInit can generate an Operator who's depth lies in [MinHeight, MaxHeight].

HoistMutation replaces an Operator by of it's operands.

The If operator.

An Initializer generates a random Operator with random operands.

The Inv operator.

A LineSearcher finds a good enough step size to a gradient descent problem. f is the function we want to minimize given a step size.

LogLoss implementes logistic loss.

MAE measures the mean absolute error (MAE).

MSE measures the mean squared error (MSE).

MacroF1 measures the global F1 score.

MacroPrecision measures the unweighted average precision across all classes. This does not take class imbalance into account.

MacroRecall measures the unweighted average recall across all classes. This does not take class imbalance into account.

The Max operator.

A Metric metricuates the performance of a predictive model. yTrue, yPred, and weights should all have the same length. If weights is nil then uniform

MicroF1 measures the global F1 score.

MicroPrecision measures the global precision by using the total true positives and false positives.

MicroRecall measures the global recall by using the total true positives and false negatives.

The Min operator.

The Mul operator.

A Mutator takes an Operator and returns a modified version of it.

The Neg operator.

A Negative returns the negative output of a given Metric.

An Operator is a mathematical operator. It has operands that are themselves Operators.

PointMutation randomly replaces Operators.

Precision measures the fraction of times a class was correctly predicted.

A Program is a thin layer on top of an Operator.

R2 measures the coefficient of determination.

RMSE measures the root mean squared error (RMSE).

ROCAUC measures the ROC AUC score.

RampedHaldAndHalfInit randomly uses GrowInit and FullInit. If it uses FullInit then it uses a random height.

Recall measures the fraction of times a true class was predicted.

The Sin operator.

The Square operator.

The Sub operator.

SubtreeCrossover applies subtree crossover to two Operators.

SubtreeMutation selects a suboperator at random and replaces it with a new Operator.

The Var operator.

WeightedF1 measures the weighted average F1 score across all classes. This does take class imbalance into account.

WeightedPrecision measures the weighted average precision across all classes. This does take class imbalance into account.

WeightedRecall measures the weighted average recall across all classes. This does take class imbalance into account.

Custom genetic algorithm model.