# -*- coding: utf-8 -*-
# This file as well as the whole tsfresh package are licenced under the MIT licence (see the LICENCE.txt)
# Maximilian Christ (maximilianchrist.com), Blue Yonder Gmbh, 2016
"""
This module contains the main function to interact with tsfresh: extract features
"""
from __future__ import absolute_import, division
import logging
import warnings
from functools import partial
from multiprocessing import Pool
import numpy as np
import pandas as pd
from builtins import str
from multiprocessing import Pool
from functools import partial
from six.moves.queue import Queue
import logging
import pandas as pd
import numpy as np
from tqdm import tqdm
from tsfresh.feature_extraction.settings import ComprehensiveFCParameters, get_aggregate_functions, get_apply_functions
from tsfresh import defaults
from tsfresh.utilities import dataframe_functions, profiling
_logger = logging.getLogger(__name__)
def _extract_features_parallel_per_kind(kind_to_df_map,
column_id, column_value,
default_fc_parameters,
kind_to_fc_parameters=None,
chunksize=defaults.CHUNKSIZE,
n_processes=defaults.N_PROCESSES, show_warnings=defaults.SHOW_WARNINGS,
disable_progressbar=defaults.DISABLE_PROGRESSBAR,
impute_function=defaults.IMPUTE_FUNCTION):
"""
Parallelize the feature extraction per kind.
:param kind_to_df_map: The time series to compute the features for in our internal format
:type kind_to_df_map: dict of pandas.DataFrame
:param column_id: The name of the id column to group by.
:type column_id: str
:param column_value: The name for the column keeping the value itself.
:type column_value: str
:param default_fc_parameters: mapping from feature calculator names to parameters. Only those names
which are keys in this dict will be calculated. See the class:`ComprehensiveFCParameters` for
more information.
:type default_fc_parameters: dict
:param kind_to_fc_parameters: mapping from kind names to objects of the same type as the ones for
default_fc_parameters. If you put a kind as a key here, the fc_parameters
object (which is the value), will be used instead of the default_fc_parameters.
:type kind_to_fc_parameters: dict
:param chunksize: The size of one chunk for the parallelisation
:type chunksize: None or int
:param n_processes: The number of processes to use for parallelisation.
:type n_processes: int
:param: show_warnings: Show warnings during the feature extraction (needed for debugging of calculators).
:type show_warnings: bool
:param disable_progressbar: Do not show a progressbar while doing the calculation.
:type disable_progressbar: bool
:param impute_function: None, if no imputing should happen or the function to call for imputing.
:type impute_function: None or function
:return: The (maybe imputed) DataFrame containing extracted features.
:rtype: pandas.DataFrame
"""
partial_extract_features_for_one_time_series = partial(_extract_features_for_one_time_series,
column_id=column_id,
column_value=column_value,
default_fc_parameters=default_fc_parameters,
kind_to_fc_parameters=kind_to_fc_parameters,
show_warnings=show_warnings)
pool = Pool(n_processes)
if not chunksize:
chunksize = _calculate_best_chunksize(kind_to_df_map, n_processes)
total_number_of_expected_results = len(kind_to_df_map)
extracted_features = tqdm(pool.imap_unordered(partial_extract_features_for_one_time_series, kind_to_df_map.items(),
chunksize=chunksize), total=total_number_of_expected_results,
desc="Feature Extraction", disable=disable_progressbar)
pool.close()
# Concatenate all partial results
result = pd.concat(extracted_features, axis=1, join='outer').astype(np.float64)
# Impute the result if requested
if impute_function is not None:
impute_function(result)
pool.join()
return result
def _extract_features_parallel_per_sample(kind_to_df_map,
column_id, column_value,
default_fc_parameters,
kind_to_fc_parameters=None,
chunksize=defaults.CHUNKSIZE,
n_processes=defaults.N_PROCESSES, show_warnings=defaults.SHOW_WARNINGS,
disable_progressbar=defaults.DISABLE_PROGRESSBAR,
impute_function=defaults.IMPUTE_FUNCTION):
"""
Parallelize the feature extraction per kind and per sample.
As the splitting of the dataframes per kind along column_id is quite costly, we settled for an async map in this
function. The result objects are temporarily stored in a fifo queue from which they can be retrieved in order
of submission.
:param kind_to_df_map: The time series to compute the features for in our internal format
:type kind_to_df_map: dict of pandas.DataFrame
:param column_id: The name of the id column to group by.
:type column_id: str
:param column_value: The name for the column keeping the value itself.
:type column_value: str
:param default_fc_parameters: mapping from feature calculator names to parameters. Only those names
which are keys in this dict will be calculated. See the class:`ComprehensiveFCParameters` for
more information.
:type default_fc_parameters: dict
:param kind_to_fc_parameters: mapping from kind names to objects of the same type as the ones for
default_fc_parameters. If you put a kind as a key here, the fc_parameters
object (which is the value), will be used instead of the default_fc_parameters.
:type kind_to_fc_parameters: dict
:param chunksize: The size of one chunk for the parallelisation
:type chunksize: None or int
:param n_processes: The number of processes to use for parallelisation.
:type n_processes: int
:param: show_warnings: Show warnings during the feature extraction (needed for debugging of calculators).
:type show_warnings: bool
:param disable_progressbar: Do not show a progressbar while doing the calculation.
:type disable_progressbar: bool
:param impute_function: None, if no imputing should happen or the function to call for imputing.
:type impute_function: None or function
:return: The (maybe imputed) DataFrame containing extracted features.
:rtype: pandas.DataFrame
"""
partial_extract_features_for_one_time_series = partial(_extract_features_for_one_time_series,
column_id=column_id,
column_value=column_value,
default_fc_parameters=default_fc_parameters,
kind_to_fc_parameters=kind_to_fc_parameters,
show_warnings=show_warnings)
pool = Pool(n_processes)
total_number_of_expected_results = 0
# Submit map jobs per kind per sample
results_fifo = Queue()
for kind, df_kind in kind_to_df_map.items():
df_grouped_by_id = df_kind.groupby(column_id)
total_number_of_expected_results += len(df_grouped_by_id)
if not chunksize:
chunksize = _calculate_best_chunksize(df_grouped_by_id, n_processes)
results_fifo.put(
pool.imap_unordered(
partial_extract_features_for_one_time_series,
[(kind, df_group) for _, df_group in df_grouped_by_id],
chunksize=chunksize
)
)
pool.close()
# Wait for the jobs to complete and concatenate the partial results
dfs_per_kind = []
# Do this all with a progress bar
with tqdm(total=total_number_of_expected_results, desc="Feature Extraction",
disable=disable_progressbar) as progress_bar:
# We need some sort of measure, when a new result is there. So we wrap the
# map_results into another iterable which updates the progress bar each time,
# a new result is there
def iterable_with_tqdm_update(queue, progress_bar):
for element in queue:
progress_bar.update(1)
yield element
result = pd.DataFrame()
while not results_fifo.empty():
map_result = results_fifo.get()
dfs_kind = iterable_with_tqdm_update(map_result, progress_bar)
df_tmp = pd.concat(dfs_kind, axis=0).astype(np.float64)
# Impute the result if requested
if impute_function is not None:
impute_function(df_tmp)
result = pd.concat([result, df_tmp], axis=1).astype(np.float64)
pool.join()
return result
def _calculate_best_chunksize(iterable_list, n_processes):
"""
Helper function to calculate the best chunksize for a given number of elements to calculate.
The formula is more or less an empirical result.
:param iterable_list: A list which defines how many calculations there need to be.
:param n_processes: The number of processes that will be used in the calculation.
:return: The chunksize which should be used.
TODO: Investigate which is the best chunk size for different settings.
"""
chunksize, extra = divmod(len(iterable_list), n_processes * 5)
if extra:
chunksize += 1
return chunksize
def _extract_features_for_one_time_series(prefix_and_dataframe, column_id, column_value,
default_fc_parameters,
kind_to_fc_parameters=None,
show_warnings=defaults.SHOW_WARNINGS):
"""
Extract time series features for a given data frame based on the passed settings.
This is an internal function, please use the extract_features function.
The `dataframe` is expected to have at least two columns: column_id and column_value. The data is grouped together
by their column_id value and the time series features are calculated independently for each of the groups.
As a result, the function returns a :class:`pandas.DataFrame` with the ids as an index and as many columns as there
were features calculated.
To distinguish the features from others, all the calculated columns are given the prefix passed in by column_prefix.
For example, if you pass in a `dataframe` of shape
+====+=======+=====+
| id | value | ... |
+====+=======+=====+
| A | 1 | ... |
+----+-------+-----+
| A | 2 | ... |
+----+-------+-----+
| A | 3 | ... |
+----+-------+-----+
| B | 1 | ... |
+----+-------+-----+
| B | 2 | ... |
+----+-------+-----+
| B | 3 | ... |
+----+-------+-----+
with `column_id="id"`, `column_value="value"` and `column_prefix="prefix"` the resulting :class:`pandas.DataFrame`
will have shape
+=======+==================+==================+=====+==================+
| Index | prefix_feature_1 | prefix_feature_2 | ... | prefix_feature_N |
+=======+==================+==================+=====+==================+
| A | ... | ... | ... | ... |
+-------+------------------+------------------+-----+------------------+
| B | ... | ... | ... | ... |
+-------+------------------+------------------+-----+------------------+
where N is the number of features that were calculated. Which features are calculated is controlled by the
passed settings instance (see :class:`~tsfresh.feature_extraction.settings.ComprehensiveFCParameters` for a list of
all possible features to calculate).
The parameter `dataframe` is not allowed to have any NaN value in it. It is possible to have different numbers
of values for different ids.
:param prefix_and_dataframe: Tuple of column_prefix and dataframe
column_prefix is the string that each extracted feature will be prefixed with (for better separation)
dataframe with at least the columns column_id and column_value to extract the time
series features for.
:type prefix_and_dataframe: (str, DataFrame)
:param column_id: The name of the column with the ids.
:type column_id: str
:param column_value: The name of the column with the values.
:type column_value: str
:param default_fc_parameters: mapping from feature calculator names to parameters. Only those names
which are keys in this dict will be calculated. See the class:`ComprehensiveFCParameters` for
more information.
:type default_fc_parameters: dict
:param kind_to_fc_parameters: mapping from kind names to objects of the same type as the ones for
default_fc_parameters. If you put a kind as a key here, the fc_parameters
object (which is the value), will be used instead of the default_fc_parameters.
:type kind_to_fc_parameters: dict
:param show_warnings: Show warnings during the feature extraction (needed for debugging of calculators).
:type show_warnings: bool
:return: A dataframe with the extracted features as the columns (prefixed with column_prefix) and as many
rows as their are unique values in the id column.
"""
if kind_to_fc_parameters is None:
kind_to_fc_parameters = {}
column_prefix, dataframe = prefix_and_dataframe
column_prefix = str(column_prefix)
# Ensure features are calculated on float64
dataframe[column_value] = dataframe[column_value].astype(np.float64)
# If there are no special settings for this column_prefix, use the default ones.
if column_prefix in kind_to_fc_parameters:
fc_parameters = kind_to_fc_parameters[column_prefix]
else:
fc_parameters = default_fc_parameters
with warnings.catch_warnings():
if not show_warnings:
warnings.simplefilter("ignore")
else:
warnings.simplefilter("default")
# Calculate the aggregation functions
column_name_to_aggregate_function = get_aggregate_functions(fc_parameters, column_prefix)
if column_name_to_aggregate_function:
extracted_features = dataframe.groupby(column_id)[column_value].aggregate(column_name_to_aggregate_function)
else:
extracted_features = pd.DataFrame(index=dataframe[column_id].unique())
# Calculate the apply functions
apply_functions = get_apply_functions(fc_parameters, column_prefix)
if apply_functions:
list_of_extracted_feature_dataframes = [extracted_features]
for apply_function, kwargs in apply_functions:
current_result = dataframe.groupby(column_id)[column_value].apply(apply_function, **kwargs).unstack()
if len(current_result) > 0:
list_of_extracted_feature_dataframes.append(current_result)
if len(list_of_extracted_feature_dataframes) > 0:
extracted_features = pd.concat(list_of_extracted_feature_dataframes, axis=1,
join_axes=[extracted_features.index])
return extracted_features