Search in the document preview
Connexions module: m22974 1
Circular Correlation Coefficient ∗
Evaristo Rojas Mayoral
This work is produced by The Connexions Project and licensed under the
Creative Commons Attribution License †
A Circular Correlation Coecient (CCC) measures the association between two angular variables.
A correlation coecient is a number that describes some characteristics of the relationship between two variables. There are several correlation coecients, each used to analyze a particular data type.
A Circular Correlation Coecient (CCC) measures the association between two angular variables. Angular variables, whose values are directions, are expressed in radians (or degrees) and they are character- ized by α = α+m (2π) for any direction α, where m is any integer.
For example, to study the relationship between the direction of a marine current and the direction of the wind on the sea surface at some point, it would be convenient to use a CCC.
The CCC proposed by Fisher and Lee (1983) is dened as:
∑n j=i+1 sin (ai − aj) sin (bi − bj)√∑n−1
∑n j=i+1 sin
2 (ai − aj) ∑n−1
∑n j=i+1 sin
2 (bi − bj) (1)
where r is the CCC, n is the number of data points, a is a angular variable and b is another angular variable. The absolute value of r takes on values within the interval from zero to one, where zero indicates that
there is no relationship between the variables and one represents the strongest association possible. For positive r the relationship is of the type a = b + γ, while for a negative r, the association has the form a = −b+ γ for a xed, arbitrary direction γ.
If you need to analyze the relationship between two angular variables, I recommend that you use this CCC, which is easily calculated using my circcorrcoef MATLAB function.
• Download circcorrcoef from MATLAB File Exchange1
• FISHER, N. & LEE, A. (1983). A correlation coecient for circular data. Biometrika. 70 (2), 327-332. DOI: 10.1093/biomet/70.2.3272
∗Version 1.3: May 31, 2009 12:21 am GMT-5 †http://creativecommons.org/licenses/by/3.0/ 1http://www.mathworks.com/matlabcentral/leexchange/23817 2http://dx.doi.org/10.1093/biomet/70.2.327