Ranked Correlation
Ranked correlation has one target variable (
Primary Variable
) and searches every other variable in the target variable's dataset (unless a select set of variables is selected on plot generation) and ranks each variable from first to last on what has the highest correlation to the Primary Variable
. The Correlation coefficient is displayed (the same number as shown on the Correlation plot unless ranked correlation is searched over a time range Offset
). This is the primary plotting option. The default is Zero
-time offset (as on Correlation plot).If desired, a range of rows can be searched and the data over the set
Search
range (and step size Interval
) is mapped and presented from highest to lowest absolute correlation coefficient along with the identified maximum row shift (secondary relative to primary variable) plus the number of rows available and used to calculate the shifted correlation coefficient (caution as sometimes high correlation coefficients at the top of the list represent very few patterns). You may adjust limits on the range of shifts in rows to search. Note that the default Search ranges assume that the primary variable is a later result of the secondary variables (Start
-20, End
0), but this may be edited as required (e.g. if the secondary variables result from changes in the primary variable change the search limit orientation, e.g. Start
: 0, End
20).The most common
Correlation Type
for this display is a Linear
correlation coefficient. A Nonlinear
correlation option is available in the selection bar. Note that the Compute
button must be selected to update calculations for correlation if the Correlation Type
or the Offset
Search
range is adjusted.The primary statistics options (Plot, Statistics or Export) are available on selecting variables. Thus the only way to adjust the Primary Variable is to select a new Ranked Correlation under the Statistics option.
Provide Feedback