feature_types

Maxent feature types to fit. Must be in string "lqphta" or list ["linear", "quadratic", "product", "hinge", "threshold", "auto"].

tau

Maxent prevalence value for scaling logistic output transform: maxent model transformation type. Select from ["raw", "logistic", "cloglog"].

clamp

Set features to min/max range from training during prediction.

scorer

Sklearn scoring function for model training.

beta_multiplier

Scaler for all regularization parameters. Higher values drop more coeffiecients.

beta_lqp

Linear, quadratic and product feature regularization scaler.

beta_hinge

Hinge feature regularization scaler.

beta_threshold

Threshold feature regularization scaler.

beta_categorical

Categorical feature regularization scaler.

n_hinge_features

The number of hinge features to fit in feature transformation.

n_threshold_features

The number of thresholds to fit in feature transformation.

convergence_tolerance

Model convergence tolerance level.

use_lambdas

Guide for which model lambdas to select (either "best" or "last").

n_lambdas

Number of lamba values to fit models with.

class_weights

Strategy for weighting presence samples. Pass "balanced" to compute the ratio based on sample frequency or pass a float for the presence:background weight ratio R maxnet package uses a value of 100 as default. Set to None to ignore.

n_cpus

Threads to use during model training.

use_sklearn

Force using "sklearn" for fitting logistic regression. Turned off by default to use "glmnet" for fitting. This feature was turned on to support Windows users install the package without a fortran compiler.