Math Functions > Statistics Functions


Linear Regression

The Linear Regression function computes the linear regression and other correlation coefficients between two time series data sets.  Values in the primary time series data set and the second time series data set are matched by time to form data pairs for the correlation analysis.   Missing values are ignored.  Times for the two time series data sets must match exactly.  The data sets may be either regular or irregular interval time series data.

 

The correlations statistics computed by the function are:

 

Number of Valid Values

Regression Constant

Regression Coefficient

Determination Coefficient

Standard Error of Regression

Adjusted Determination Coefficient

Adjusted Standard Error of Regression

 

The primary time series data set forms the values of the independent variable (x-values), while values of the second time series data set comprise the dependent variable (y-values).  The linear regression coefficients express how values in the second data set can be derived from values in the primary data set:

 

TS2(t) = a + b * TS1(t)

 

where “a” is the regression constant and “b” the regression coefficient.

 

Linear Regression Statistics

 

To compute the linear regression and correlation coefficients between two time series data sets:

  1. Choose the Statistics tab of the Math Functions screen and select the Linear Regression type, as shown in the figure above.
  2. From the Selected Data Set list, select the primary time series data set for analysis.  Time series values from this data set form the independent variable (x-values) of the correlation analysis.
  3. From the Dependent Data Set list, select the second time series data set for analysis.  Time series values from this data set will comprise the dependent variable (y-values) of the correlation analysis.

    The correlation statistics are computed automatically once the two data sets are selected.