foundryts.functions.linear_regression

foundryts.functions.linear_regression(include_intercept=True, time_unit='ns', start=None, end=None)

Returns a function that performs linear regression on a single time series.

Linear regression finds the parameters of the best-fit line over points of the input time series. Linear regression is expressed as y = Ax + B, where A is the slope of the line and B is the y-intercept. The returned function will provide the parameters A and B.

Linear regression is useful when you need to identify and quantify a linear trend in your time series data.

Parameters:
- time_unit (str , optional) – The time unit of the coefficients, must be one of “s”, “ms”, “us”, “ns” (default is “ns”).
- start (str | int | datetime.datetime , optional) – Starting point (inclusive) of the time series for computing the linear regression.
- end (str | int | datetime.datetime , optional) – End point (exclusive) of the time series for computing the linear regression.
Returns: A function that accepts a single time series and returns parameters for the best-fit line for the points in the time series using linear regression.
Return type: (FunctionNode) -> SummarizerNode

Dataframe schema

Column name	Type	Description
max_bounds.first_value	float	Maximum value of the slope (A) in `y=Ax+B`.
max_bounds.second_value	float	Maximum value of the intercept (B) in `y=Ax+B`.
min_bounds.first_value	float	Minimum value of the slope (A) in `y=Ax+B`.
min_bounds.second_value	float	Minimum value of the intercept (B) in `y=Ax+B`.
regression_fit_function. linear_regression_fit. slope	float	Parameter ‘A’ (slope) of the linear regression fit in `y=Ax+B`.
regression_fit_function. linear_regression_fit. intercept	float	Parameter ‘B’ (intercept) of the linear regression fit in `y=Ax+B`.
regression_fit_function. linear_regression_fit. statistics.rsquared	float	R-squared value indicating the goodness of fit of the linear regression.

Examples

Copied!1
2
3
4
5
6
7
8
9
10
>>> series = F.points(
...     (10, 6.0), (20, 12.0), (30, 24.0), (40, 48.0), (50, 96.0), name="series"
... )
>>> series.to_pandas()
                      timestamp  value
0 1970-01-01 00:00:00.000000010    6.0
1 1970-01-01 00:00:00.000000020   12.0
2 1970-01-01 00:00:00.000000030   24.0
3 1970-01-01 00:00:00.000000040   48.0
4 1970-01-01 00:00:00.000000050   96.0

Copied!1
2
3
4
>>> lin_regr = F.linear_regression()(series)
>>> lin_regr.to_pandas()
   max_bounds.first_value  max_bounds.second_value  min_bounds.first_value  min_bounds.second_value  regression_fit_function.linear_regression_fit.intercept  regression_fit_function.linear_regression_fit.slope  regression_fit_function.linear_regression_fit.statistics.rsquared
0                    50.0                     96.0                    10.0                      6.0                                              -27.6                                                     2.16                                             0.870968