.rst .pdf

repository open issue suggest edit

Contents

zoo.automl.common package

Submodules

zoo.automl.common.metrics module

zoo.automl.common.metrics.Accuracy(y_true, y_pred, multioutput=None)

class zoo.automl.common.metrics.Evaluator

Bases: object

Evaluate metrics for y_true and y_pred

static check_metric(metric)

static evaluate(metric, y_true, y_pred, multioutput='raw_values')

metrics_func = {'accuracy': <function Accuracy>, 'mae': <function mean_absolute_error>, 'mape': <function MAPE>, 'mdape': <function MDAPE>, 'me': <function ME>, 'mpe': <function MPE>, 'mse': <function mean_squared_error>, 'msle': <function mean_squared_log_error>, 'mspe': <function MSPE>, 'r2': <function r2_score>, 'rmse': <function RMSE>, 'smape': <function sMAPE>, 'smdape': <function sMDAPE>}

zoo.automl.common.metrics.MAPE(y_true, y_pred, multioutput='raw_values')

calculate mean absolute percentage error (MAPE). <math>mbox{M} =

rac{100%}{n}sum_{t=1}^n left| rac{A_t-F_t}{A_t} ight|, </math>

param y_true

array-like of shape = (n_samples) or (n_samples, n_outputs) Ground truth (correct) target values.

param y_pred

array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

param multioutput

string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.MDAPE(y_true, y_pred, multioutput='raw_values')

calculate Median Absolute Percentage Error (MDAPE). :param y_true: array-like of shape = (n_samples) or (n_samples, n_outputs)

Ground truth (correct) target values.

Parameters

• y_pred – array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

• multioutput – string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.ME(y_true, y_pred, multioutput='raw_values')

calculate Mean Error (ME). :param y_true: array-like of shape = (n_samples) or (n_samples, n_outputs)

Ground truth (correct) target values.

Parameters

• y_pred – array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

• multioutput – string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.MPE(y_true, y_pred, multioutput='raw_values')

calculate mean percentage error (MPE). <math> ext{MPE} =

rac{100%}{n}sum_{t=1}^n rac{a_t-f_t}{a_t} </math>

param y_true

array-like of shape = (n_samples) or (n_samples, n_outputs) Ground truth (correct) target values.

param y_pred

array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

param multioutput

string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.MSPE(y_true, y_pred, multioutput='raw_values')

calculate mean squared percentage error (MSPE). <math>operatorname{MSPE}(L)=operatorname{E} left[left( g(x_i)-widehat{g}(x_i)

ight)^2 ight].</math>

param y_true

array-like of shape = (n_samples) or (n_samples, n_outputs) Ground truth (correct) target values.

param y_pred

array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

param multioutput

string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.RMSE(y_true, y_pred, multioutput='raw_values')

calculate square root of the mean squared error (RMSE). :param y_true: array-like of shape = (n_samples) or (n_samples, n_outputs)

Ground truth (correct) target values.

Parameters

• y_pred – array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

• multioutput – string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.check_input(y_true, y_pred, multioutput)

zoo.automl.common.metrics.sMAPE(y_true, y_pred, multioutput='raw_values')

calculate Symmetric mean absolute percentage error (sMAPE). <math> ext{SMAPE} =

rac{100%}{n} sum_{t=1}^n rac{|F_t-A_t|}{|A_t|+|F_t|}</math>

param y_true

array-like of shape = (n_samples) or (n_samples, n_outputs) Ground truth (correct) target values.

param y_pred

array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

param multioutput

string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.metrics.sMDAPE(y_true, y_pred, multioutput='raw_values')

calculate Symmetric Median Absolute Percentage Error (sMDAPE). :param y_true: array-like of shape = (n_samples) or (n_samples, n_outputs)

Ground truth (correct) target values.

Parameters

• y_pred – array-like of shape = (n_samples) or (n_samples, n_outputs) Estimated target values.

• multioutput – string in [‘raw_values’, ‘uniform_average’]

:return:float or ndarray of floats

A non-negative floating point value (the best value is 0.0), or an array of floating point values, one for each individual target.

zoo.automl.common.parameters module

zoo.automl.common.util module

class zoo.automl.common.util.NumpyEncoder(*, skipkeys=False, ensure_ascii=True, check_circular=True, allow_nan=True, sort_keys=False, indent=None, separators=None, default=None)

Bases: json.encoder.JSONEncoder

convert numpy array to list for JSON serialize

default(obj)

Implement this method in a subclass such that it returns a serializable object for o, or calls the base implementation (to raise a TypeError).

For example, to support arbitrary iterators, you could implement default like this:

def default(self, o):
    try:
        iterable = iter(o)
    except TypeError:
        pass
    else:
        return list(iterable)
    # Let the base class default method raise the TypeError
    return JSONEncoder.default(self, o)

zoo.automl.common.util.convert_bayes_configs(config)

zoo.automl.common.util.get_remote_list(dir_in)

zoo.automl.common.util.load_config(file_path)

zoo.automl.common.util.process(cmd)

zoo.automl.common.util.restore(file, feature_transformers=None, model=None, config=None)

zoo.automl.common.util.restore_hdfs(model_path, remote_dir, feature_transformers=None, model=None, config=None)

zoo.automl.common.util.restore_zip(file, feature_transformers=None, model=None, config=None)

zoo.automl.common.util.save(file_path, feature_transformers=None, model=None, config=None)

zoo.automl.common.util.save_config(file_path, config, replace=False)

Parameters

• file_path – the file path of config to be saved.

• config – dict. The config to be saved

• replace – whether to replace if the config file already existed.

Returns

zoo.automl.common.util.save_zip(file, feature_transformers=None, model=None, config=None)

zoo.automl.common.util.train_val_test_split(df, val_ratio=0, test_ratio=0.1, look_back=0, horizon=1)

split input dataframe into train_df, val_df and test_df according to split ratio. The dataframe is splitted in its originally order in timeline. e.g. |......... train_df(80%) ........ | ... val_df(10%) ...| …test_df(10%)…| :param df: dataframe to be splitted :param val_ratio: validation ratio :param test_ratio: test ratio :param look_back: the length to look back :param horizon: num of steps to look forward :return:

zoo.automl.common.util.upload_ppl_hdfs(upload_dir, ckpt_name)

Module contents

zoo.automl package zoo.automl.config package