statistics.models
class GAD7Labeler
Section titled “class GAD7Labeler”Bases: object
__init__()
Section titled “__init__()”fit(X, y=None)
Section titled “fit(X, y=None)”Method kept for consistency with the scikit-learn API. But in this case, will simply call the transform method since no actual model gets fitted here.
Needs to contain only 7 columns representing the GAD-7 items. Should be in the form of a DataFrame with shape (n_samples, 7).
Parameters
Section titled “Parameters”X : Input features (n_samples, 7) with only the 7 GAD-7 items as columns and subject as rows.
y : Target variable. Not used in this model. Keeps the API consistent.
Returns
Section titled “Returns”pd.DataFrame : Transformed features.
transform(X)
Section titled “transform(X)”Transform the input features to assign the label based on the GAD-7 scoring.
Needs to contain only 7 columns representing the GAD-7 items. Should be in the form of a DataFrame with shape (n_samples, 7).
Final labels will be either:
- Not anxious
- Mild
- Moderate
- Severe
Parameters
Section titled “Parameters”X : Input features.
Returns
Section titled “Returns”pd.DataFrame : Transformed features.
class PHQ9Labeler
Section titled “class PHQ9Labeler”Bases: object
__init__()
Section titled “__init__()”fit(X, y=None)
Section titled “fit(X, y=None)”Method kept for consistency with the scikit-learn API. But in this case, will simply call the transform method since no actual model gets fitted here.
Needs to contain only 9 columns representing the ordered PHQ-9 items. Should be in the form of a DataFrame with shape (n_samples, 9).
Parameters
Section titled “Parameters”X : Input features (n_samples, 9) with only the 9 PHQ-9 items as columns and subject as rows.
y : Target variable. Not used in this model. Keeps the API consistent.
Returns
Section titled “Returns”pd.DataFrame : Transformed features.
transform(X)
Section titled “transform(X)”Transform the input features to assign the label based on the fuzzy weighting of the PHQ-9 items. Needs to contain only 9 columns representing the ordered PHQ-9 items. Should be in the form of a DataFrame with shape (n_samples, 9).
Final labels will be either:
- Not depressed
- Mild
- Moderate
- Mod-Severe
- Severe
Parameters
Section titled “Parameters”X : Input features.
Returns
Section titled “Returns”pd.DataFrame : Transformed features.
class Penalty(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)
Section titled “class Penalty(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)”Bases: StrEnum, Enum
elasticnet = ‘elasticnet’
Section titled “elasticnet = ‘elasticnet’”l1 = ‘l1’
Section titled “l1 = ‘l1’”l2 = ‘l2’
Section titled “l2 = ‘l2’”class ScoringMethod(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)
Section titled “class ScoringMethod(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)”Bases: StrEnum, Enum
accuracy = ‘accuracy’
Section titled “accuracy = ‘accuracy’”adjusted_mutual_info_score = ‘adjusted_mutual_info_score’
Section titled “adjusted_mutual_info_score = ‘adjusted_mutual_info_score’”adjusted_rand_score = ‘adjusted_rand_score’
Section titled “adjusted_rand_score = ‘adjusted_rand_score’”average_precision = ‘average_precision’
Section titled “average_precision = ‘average_precision’”balanced_accuracy = ‘balanced_accuracy’
Section titled “balanced_accuracy = ‘balanced_accuracy’”completeness_score = ‘completeness_score’
Section titled “completeness_score = ‘completeness_score’”d2_absolute_error_score = ‘d2_absolute_error_score’
Section titled “d2_absolute_error_score = ‘d2_absolute_error_score’”d2_pinball_score = ‘d2_pinball_score’
Section titled “d2_pinball_score = ‘d2_pinball_score’”d2_tweedie_score = ‘d2_tweedie_score’
Section titled “d2_tweedie_score = ‘d2_tweedie_score’”explained_variance = ‘explained_variance’
Section titled “explained_variance = ‘explained_variance’”f1 = ‘f1’
Section titled “f1 = ‘f1’”f1_macro = ‘f1_macro’
Section titled “f1_macro = ‘f1_macro’”f1_micro = ‘f1_micro’
Section titled “f1_micro = ‘f1_micro’”f1_samples = ‘f1_samples’
Section titled “f1_samples = ‘f1_samples’”f1_weighted = ‘f1_weighted’
Section titled “f1_weighted = ‘f1_weighted’”fowlkes_mallows_score = ‘fowlkes_mallows_score’
Section titled “fowlkes_mallows_score = ‘fowlkes_mallows_score’”homogeneity_score = ‘homogeneity_score’
Section titled “homogeneity_score = ‘homogeneity_score’”jaccard = ‘jaccard’
Section titled “jaccard = ‘jaccard’”max_error = ‘max_error’
Section titled “max_error = ‘max_error’”mutual_info_score = ‘mutual_info_score’
Section titled “mutual_info_score = ‘mutual_info_score’”neg_brier_score = ‘neg_brier_score’
Section titled “neg_brier_score = ‘neg_brier_score’”neg_log_loss = ‘neg_log_loss’
Section titled “neg_log_loss = ‘neg_log_loss’”neg_mean_absolute_error = ‘neg_mean_absolute_error’
Section titled “neg_mean_absolute_error = ‘neg_mean_absolute_error’”neg_mean_absolute_percentage_error = ‘neg_mean_absolute_percentage_error’
Section titled “neg_mean_absolute_percentage_error = ‘neg_mean_absolute_percentage_error’”neg_mean_gamma_deviance = ‘neg_mean_gamma_deviance’
Section titled “neg_mean_gamma_deviance = ‘neg_mean_gamma_deviance’”neg_mean_poisson_deviance = ‘neg_mean_poisson_deviance’
Section titled “neg_mean_poisson_deviance = ‘neg_mean_poisson_deviance’”neg_mean_squared_error = ‘neg_mean_squared_error’
Section titled “neg_mean_squared_error = ‘neg_mean_squared_error’”neg_mean_squared_log_error = ‘neg_mean_squared_log_error’
Section titled “neg_mean_squared_log_error = ‘neg_mean_squared_log_error’”neg_median_absolute_error = ‘neg_median_absolute_error’
Section titled “neg_median_absolute_error = ‘neg_median_absolute_error’”neg_root_mean_squared_error = ‘neg_root_mean_squared_error’
Section titled “neg_root_mean_squared_error = ‘neg_root_mean_squared_error’”normalized_mutual_info_score = ‘normalized_mutual_info_score’
Section titled “normalized_mutual_info_score = ‘normalized_mutual_info_score’”precision = ‘precision’
Section titled “precision = ‘precision’”r2 = ‘r2’
Section titled “r2 = ‘r2’”rand_score = ‘rand_score’
Section titled “rand_score = ‘rand_score’”recall = ‘recall’
Section titled “recall = ‘recall’”roc_auc = ‘roc_auc’
Section titled “roc_auc = ‘roc_auc’”roc_auc_ovo = ‘roc_auc_ovo’
Section titled “roc_auc_ovo = ‘roc_auc_ovo’”roc_auc_ovo_weighted = ‘roc_auc_ovo_weighted’
Section titled “roc_auc_ovo_weighted = ‘roc_auc_ovo_weighted’”roc_auc_ovr = ‘roc_auc_ovr’
Section titled “roc_auc_ovr = ‘roc_auc_ovr’”roc_auc_ovr_weighted = ‘roc_auc_ovr_weighted’
Section titled “roc_auc_ovr_weighted = ‘roc_auc_ovr_weighted’”top_k_accuracy = ‘top_k_accuracy’
Section titled “top_k_accuracy = ‘top_k_accuracy’”v_measure_score = ‘v_measure_score’
Section titled “v_measure_score = ‘v_measure_score’”class Solver(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)
Section titled “class Solver(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)”Bases: StrEnum, Enum
lbfgs = ‘lbfgs’
Section titled “lbfgs = ‘lbfgs’”liblinear = ‘liblinear’
Section titled “liblinear = ‘liblinear’”newton_cg = ‘newton-cg’
Section titled “newton_cg = ‘newton-cg’”newton_cholesky = ‘newton-cholesky’
Section titled “newton_cholesky = ‘newton-cholesky’”sag = ‘sag’
Section titled “sag = ‘sag’”saga = ‘saga’
Section titled “saga = ‘saga’”permutation_testing(estimator, X, Y, binary=False, nb_permutations=1000, scoring=‘r2’, splits=10, processes=1, verbose=False)
Section titled “permutation_testing(estimator, X, Y, binary=False, nb_permutations=1000, scoring=‘r2’, splits=10, processes=1, verbose=False)”Function to perform permutation testing on a model.
Parameters
Section titled “Parameters”estimator : Model to use.
X : Dataframe containing the predictor variables.
Y : Dataframe containing the dependent variables.
binary : If the dependent variable is binary. Defaults to False.
nb_permutations : Number of iterations to perform. Defaults to 1000.
scoring : Scoring method to use. Defaults to ‘r2’.
splits : Number of fold to use in cross-validation. Defaults to 10.
processes : Number of cpus to use during processing. Defaults to 1.
verbose : Verbose mode. Defaults to False.
Returns
Section titled “Returns”mod : Model.
score : Score for the model.
coef : Coefficients for the model.
perm_score : Scores for the permutation testing.
score_pvalue : P-value for the model.
perm_coef : Coefficients for the permutation testing.
coef_pvalue : P-value for the coefficients.
plsr_cv(X, Y, nb_comp, max_iter=1000, splits=10, processes=1, verbose=False)
Section titled “plsr_cv(X, Y, nb_comp, max_iter=1000, splits=10, processes=1, verbose=False)”Function to perform a PLSR model with cross-validation between a set of predictor and dependent variables.
Parameters
Section titled “Parameters”X : Dataframe containing the predictor variables.
Y : Dataframe containing the dependent variables.
nb_comp : Number of components to use.
max_iter : Maximum number of iterations. Defaults to 1000.
splits : Number of fold to use in cross-validation. Defaults to 10.
processes : Number of cpus to use during processing. Defaults to 1.
verbose : Verbose mode. Defaults to False.
Returns
Section titled “Returns”plsr : PLSR model.
mse : List of mean squared errors.
score_c : R2 score for the model.
score_cv : R2 score for the cross-validation.
rscore : Square root of the R2 score.
mse_c : Mean squared error for the model.
mse_cv : Mean squared error for the cross-validation.