sensortoolkit.calculate._regression_stats

This module computes statistical metrics for the Ordinary Least Squares linear regression between sensor and FRM/FEM datasets (FRM/FEM as the dependent variable along the x-axis and sensor data as the independent variable along the y-axis).

Bias and linearity

U.S. EPA’s Performance Targets Reports recommend using a linear regression model, relating sensor and FRM/FEM measurements, to determine the magnitude of bias and linearity. The regression model takes the form

\[y = mx + b\]

where

\(y\) = sensor measurements

\(x\) = FRM/FEM measurements

\(m\) = regression slope

\(y\) = regression intercept

The slope \(m\) and intercept \(y\) indicate the degree of systematic bias between sensor and reference measurements.

Linearity is measured via the coefficient of determination (\(R^2\)).


@Author:
Samuel Frederick, NSSC Contractor (ORAU)
U.S. EPA / ORD / CEMM / AMCD / SFSB
Created:

Tue Mar 3 13:47:32 2020

Last Updated:

Tue Jul 13 11:09:13 2021

Functions

dataframe_to_csv

Save a pandas DataFrame to a comma-separated value file.

join_stats

Combine 1-hour and 24-hour regression statistics DataFrames.

regression_stats

Compute OLS regression statistics.