Help with calculating entropy of data
mnstn wrote:
Hello All, My question is not directly related to R but rather on which statistical method I should look in to for capturing the entropy of a data-set as a number. In this figure http://www.twitpic.com/18sob5 are two data sets blue and green (x-axis is time) that fluctuate between (-1,+1). Clearly, green has 4 jumps while blue has 1 (and a some?). Intuitively, green has more entropy than blue. Is there a robust statistical quantity that can capture their relative flexibilities? Additionally I am hoping this method will differentiate between two cases where both spend 50% of time in each of the states -1 and +1 but one has more jumps than the other. I am guessing the limits of that quantity are 0 (no change) and N-1 (for N time steps). Sigma( mod(value(t)-value(t-1))/2 )? I am just thinking out loud here. I have about 200 such plots and I would like to arrange them in order of their entropy. Thanks and I sincerely apologize if you feel this is not the right place to ask this question. MoonStone
Google "information theory" and "entropy," and you'll find lots of stuff. If memory serves me, Henri Theil's book, "Economics and Information Theory," has a good discussion of this kind of stuff. Marsh Feldman
View this message in context: http://n4.nabble.com/Help-with-calculating-entropy-of-data-tp1593954p1595156.html Sent from the R help mailing list archive at Nabble.com.