mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ey-Chih chow <eyc...@gmail.com>
Subject issues with the partial implementation of random forest
Date Mon, 14 Jan 2013 20:15:50 GMT
Hi,

I tried to use Mahout PartialBuilder to build a random forest with our training data.  In
our data, the values first two attributes are all 0.0.  However, after sampling by Mahout,
some of the data set has values -0.0, NaN for the first two attributes respectively.  Consequently,
the class OptIgSplit throws ArrayIndexOutOfBoundException at the method computeFreauencies().
 Also the sizes of sampling datasets vary drastically.  For our case, we got sizes of 9783,
19, 3, 4, 144, 12, ….  Are these are issues of partial implementation?  Thanks.


Ey-Chih Chow
Mime
View raw message