Cramer's V statistic measures the strength of relationship between two categorical inputs.
For a categorical variable with N levels Vs a binary target,
Cramer's V = SQRT( CHI-SQUARE/N), which takes a value between o and 1
CHI-SQUARE statistic should be calcualted with the null hypothesis that the two variables are independent.
In the case of interval variables, the input interval variable is first binned or grouped. Then for each level of grouped variable vs the each level of target variable, MxN matrix is determined and for each cell the chi-square value is calculated. The sum of all these chi-square values gives the overall chi-square. Using the above formula, Cramer's V is determined.
Saturday, April 3, 2010
Saturday, March 27, 2010
Two Crows
Two Crows is a company specialized in providing data mining solutions. Browse through this website for more information...
http://www.twocrows.com/
http://www.twocrows.com/
Yahoo and SAS Execs Launch Data Mining Company
Check this out:
http://www.information-management.com/news/data_mining_management-10017526-1.html
http://www.information-management.com/news/data_mining_management-10017526-1.html
Subscribe to:
Comments (Atom)