This works for me, includes doctest code.
I would appreciate comments of any type as I am new to R and python.
The only problems are
(i) Warning messages:
1: closing unused connection 12
(c:\DOCUME~1\bdb112\LOCALS~1\Temp\RtmpbhGMMz\file6c6c10d9)
and
(ii) occasionally (about 1 in 10 or 1 in 20 times?) there is a 10
minute delay on a clustering operation that normally takes 10s of seconds!
Win 32: R 2.7.0
Python 2.5.2 (r252:60911, Feb 21 2008, 13:11:45) [MSC v.1310 32 bit
(Intel)] on win32
RPy 2.0.3
--
import pdb
from matplotlib.cbook import is_numlike
from numpy import array, shape
#from rpy2
def Robj_to_dictionary(robj, verbose=0, depth=0):
""" takes an RVector and converts it to a python dictionary
iterate over any elements at this level, making a dictionary
Some R features such as labels of columns and rows are ignored.
Also, conversion to integer is a little fudgey but should be safe
You can check the accuracy by
mClustRObj.r_repr() and comparing (can see the array dims also)
In the following example, several preparatory lines are supplied
in the test body (end of this code).
>>> mClustDic = Robj_to_dictionary(mClustRObj, verbose=verbose)
>>> print mClustDic["classification"]
[1 1 1 1 1 2 2 2 2]
>>> print mClustDic["parameters"]["mean"]
[[ 2.96117876 7.40416875]
[ 0.05886086 0.40401195]]
"""
elts = len(robj)
if verbose: print("%d elements found " % elts)
# open a blank dictionary
dic = {}
for (i,ro) in enumerate(robj):
name=robj.names[i]
if verbose>2: print("depth %d, element %d/%d is %s, type=%d" % (depth,
i, elts, name, ro.typeof))
if name != "NULL":
typ = ro.typeof
if typ==0:
value = "Null"
elif typ==13: # scalar
value = ro[0]
elif typ==14: # array
value = array(ro)
# fudgey test - look for . in string rep
if (ro.__str__().find('.') == -1 and is_numlike(value)):
value=value.astype(int)
# array(Robject-scalar) returns a an array([xxx]) - so fix
it
if (len(shape(value))==1) and len(value)==1: value=value[0]
elif typ==16: # string
value = ro[0]
elif typ==19:
value = Robj_to_dictionary(ro,depth=depth+1)
else:
print("unknown type %d, %s" % (typ, array(ro)))
value="unk"
x=0
y=1/x
dic.update({name: value})
if verbose>10: pdb.set_trace()
return(dic)
if __name__ == "__main__":
import doctest
import pdb
from numpy import arange, cos, sin, array
import rpy2.robjects as robjects
r = robjects.r
r.library('mclust')
verbose=0
nn=10
m = {'var1': robjects.FloatVector(arange(1,nn)),
'var2': robjects.FloatVector(sin(arange(1,nn)))}
datafr = robjects.r['data.frame'](**m)
robjects.globalEnv["datafr"] = datafr
if verbose>0: print(datafr.colnames())
mClustRObj=r.Mclust(r.__getitem__("datafr"),G=2) #,modelNames='EEI')
if verbose>0:
for i in arange(len(mClustRObj)):
print("%s %s" % (mClustRObj.names[i],mClustRObj[i][0]))
if verbose>2: r.plot(r.__getitem__("datafr"))
# note: the last two lines are in the doctest part.
doctest.testmod()
------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today.
Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________
rpy-list mailing list
rpy-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rpy-list
