I have rainfall data series for a particular city. With the help of statistics i would like to classify the data of rainfall into light rain , heavy and very heavy. Is it possible to know the threshold value of each classification with help of statistics.
Asked
Active
Viewed 28 times
0
-
Is there any reason to believe that there are real, latent groupings of rainfall like that? Or is this totally arbitrary? What would be the point of classifying rainfall in this manner? – gung - Reinstate Monica May 05 '16 at 13:19
-
Maybe a multinomial model would do? Or look into k-means clustering with k=3 clusters. – FisherDisinformation May 05 '16 at 13:32
1 Answers
0
The first thing you should do is visualize your data. If you create a scatter plot you may be able to see groupings of rainfall. Perhaps your graph will reveal that 4 or 5 categories of rainfall is better? After visualizing I would try clustering as @ArtificialBreeze suggested.
Brent Ferrier
- 101