Reputation: 23
I'm trying to run h2o.svd in spark cluster via sparkling water & h2o. The process went well and I could get the SVD object from h2o command but I could only see the result below.
#Exclude ID column in h2o data frame
my_svd <- h2o.svd(my_h2o_df[,2:138], nv = 10)
my_svd
Model Details:
==============
H2ODimReductionModel: svd
Model ID: SVD_model_R_1537868492645_2
Singular values:
sval1 sval2 sval3 sval4 sval5 sval6 sval7
1 80.821459 53.024006 40.153390 38.508806 36.984611 35.530345 33.960273
sval8 sval9 sval10
1 33.189426 27.904307 27.607862
NULL
Basically, in base R I can use svd
and it'll provide the result of $d
, $u
, $v
simultaneously in the model object.
Since I'm new to h2o workflow, I assume that the the result from above h2o object return only $d
only compare to base R. How could I get the matrices $u
and $v
?
My Environment
Upvotes: 1
Views: 155
Reputation: 5778
you can get your D, V, and U as follows:
# singular values (D)
my_svd@model$d
# singular vectors (U)
h2o.getFrame(my_svd@model$u_key$name)
# # singular vectors (V):
h2o.getFrame(my_svd@model$v_key$name)
Upvotes: 1