featuretools.primitives.RollingMax#

class featuretools.primitives.RollingMax(window_length=3, gap=1, min_periods=1)[source]#

Determines the maximum of entries over a given window.

Description:

Given a list of numbers and a corresponding list of datetimes, return a rolling maximum of the numeric values, starting at the row gap rows away from the current row and looking backward over the specified window (by window_length and gap).

Input datetimes should be monotonic.

Parameters:

  • window_length (int, string, optional) – Specifies the amount of data included in each window. If an integer is provided, it will correspond to a number of rows. For data with a uniform sampling frequency, for example of one day, the window_length will correspond to a period of time, in this case, 7 days for a window_length of 7. If a string is provided, it must be one of pandas’ offset alias strings (‘1D’, ‘1H’, etc), and it will indicate a length of time that each window should span. The list of available offset aliases can be found at https://pandas.pydata.org/pandas-docs/stable/user_guide/timeseries.html#offset-aliases. Defaults to 3.

  • gap (int, string, optional) – Specifies a gap backwards from each instance before the window of usable data begins. If an integer is provided, it will correspond to a number of rows. If a string is provided, it must be one of pandas’ offset alias strings (‘1D’, ‘1H’, etc), and it will indicate a length of time between a target instance and the beginning of its window. Defaults to 1.

  • min_periods (int, optional) – Minimum number of observations required for performing calculations over the window. Can only be as large as window_length when window_length is an integer. When window_length is an offset alias string, this limitation does not exist, but care should be taken to not choose a min_periods that will always be larger than the number of observations in a window. Defaults to 1.

Note

Only offset aliases with fixed frequencies can be used when defining gap and window_length. This means that aliases such as M or W cannot be used, as they can indicate different numbers of days. (‘M’, because different months have different numbers of days; ‘W’ because week will indicate a certain day of the week, like W-Wed, so that will indicate a different number of days depending on the anchoring date.)

Note

When using an offset alias to define gap, an offset alias must also be used to define window_length. This limitation does not exist when using an offset alias to define window_length. In fact, if the data has a uniform sampling frequency, it is preferable to use a numeric gap as it is more efficient.

Examples

>>> import pandas as pd
>>> rolling_max = RollingMax(window_length=3)
>>> times = pd.date_range(start='2019-01-01', freq='1min', periods=5)
>>> rolling_max(times, [4, 3, 2, 1, 0]).tolist()
[nan, 4.0, 4.0, 4.0, 3.0]

We can also control the gap before the rolling calculation.

>>> import pandas as pd
>>> rolling_max = RollingMax(window_length=3, gap=0)
>>> times = pd.date_range(start='2019-01-01', freq='1min', periods=5)
>>> rolling_max(times, [4, 3, 2, 1, 0]).tolist()
[4.0, 4.0, 4.0, 3.0, 2.0]

We can also control the minimum number of periods required for the rolling calculation.

>>> import pandas as pd
>>> rolling_max = RollingMax(window_length=3, min_periods=3, gap=0)
>>> times = pd.date_range(start='2019-01-01', freq='1min', periods=5)
>>> rolling_max(times, [4, 3, 2, 1, 0]).tolist()
[nan, nan, 4.0, 3.0, 2.0]

We can also set the window_length and gap using offset alias strings.

>>> import pandas as pd
>>> rolling_max = RollingMax(window_length='3min', gap='1min')
>>> times = pd.date_range(start='2019-01-01', freq='1min', periods=5)
>>> rolling_max(times, [4, 3, 2, 1, 0]).tolist()
[nan, 4.0, 4.0, 4.0, 3.0]
__init__(window_length=3, gap=1, min_periods=1)[source]#

Methods

__init__([window_length, gap, min_periods])

flatten_nested_input_types(input_types)

Flattens nested column schema inputs into a single list.

generate_name(base_feature_names)

generate_names(base_feature_names)

get_args_string()

get_arguments()

get_description(input_column_descriptions[, ...])

get_filepath(filename)

get_function()

Attributes

base_of

base_of_exclude

commutative

default_value

Default value this feature returns if no data found.

description_template

input_types

woodwork.ColumnSchema types of inputs

max_stack_depth

name

Name of the primitive

number_output_features

Number of columns in feature matrix associated with this feature

return_type

ColumnSchema type of return

stack_on

stack_on_exclude

stack_on_self

uses_calc_time

uses_full_dataframe