aggregations

This module contains all possible aggregations to be used with Multirel, FastProp, Mapping.

Refer to the feature learning section in the user guide for details about how these aggregations are used in the context of feature learning.

Functions

namedtuple(typename, field_names, *[, ...])

Returns a new subclass of tuple with named fields.

Attributes

Avg	Average value of a given numerical column.
Count	Number of rows in a given column.
CountDistinct	Count function with distinct clause.
CountDistinctOverCount	COUNT DISTINCT divided by COUNT.
CountMinusCountDistinct	Counts minus counts distinct.
EWMA_1s	Exponentially weighted moving average with a half-life of 1 second.
EWMA_1m	Exponentially weighted moving average with a half-life of 1 minute.
EWMA_1h	Exponentially weighted moving average with a half-life of 1 hour.
EWMA_1d	Exponentially weighted moving average with a half-life of 1 day.
EWMA_7d	Exponentially weighted moving average with a half-life of 7 days.
EWMA_30d	Exponentially weighted moving average with a half-life of 30 days.
EWMA_90d	Exponentially weighted moving average with a half-life of 90 days.
EWMA_365d	Exponentially weighted moving average with a half-life of 365 days.
EWMA_TREND_1s	Exponentially weighted trend with a half-life of 1 second.
EWMA_TREND_1m	Exponentially weighted trend with a half-life of 1 minute.
EWMA_TREND_1h	Exponentially weighted trend with a half-life of 1 hour.
EWMA_TREND_1d	Exponentially weighted trend with a half-life of 1 day.
EWMA_TREND_7d	Exponentially weighted trend with a half-life of 7 days.
EWMA_TREND_30d	Exponentially weighted trend with a half-life of 30 days.
EWMA_TREND_90d	Exponentially weighted trend with a half-life of 90 days.
EWMA_TREND_365d	Exponentially weighted trend with a half-life of 365 days.
First	First value of a given column, when ordered by the time stamp.
Kurtosis	The kurtosis of a given column.
Last	Last value of a given column, when ordered by the time stamp.
Max	Largest value of a given column.
Median	Median of a given column
Min	Smallest value of a given column.
Mode	Most frequent value of a given column.
NumMax	The number of times we observe the maximum value.
NumMin	The number of times we observe the minimum value.
Q1	The 1%-quantile.
Q5	The 5%-quantile.
Q10	The 10%-quantile.
Q25	The 25%-quantile.
Q75	The 75%-quantile.
Q90	The 90%-quantile.
Q95	The 95%-quantile.
Q99	The 99%-quantile.
Skew	Skewness of a given column.
Stddev	Standard deviation of a given column.
Sum	Total sum of a given numerical column.
TimeSinceFirstMaximum	The time difference between the first time we see the maximum value and the time stamp in the population table.
TimeSinceFirstMinimum	The time difference between the first time we see the minimum value and the time stamp in the population table.
TimeSinceLastMaximum	The time difference between the last time we see the maximum value and the time stamp in the population table.
TimeSinceLastMinimum	The time difference between the last time we see the minimum value and the time stamp in the population table.
Trend	Extracts a linear trend from a variable over time and extrapolates this trend to the current time stamp.
Var	Statistical variance of a given numerical column.
VariationCoefficient	VAR divided by MEAN.
fastprop	Set of default aggregations for FastProp and FastPropTimeseries.
mapping	Set of default aggregations for Mapping.
multirel	Set of default aggregations for Multirel.