在 python 中读取 RDa 文件作为 Pandas 数据框

Question

I have an RDa file that I created in R. I want to read this file on python as a pandas dataframe. I have the following code to do the same:

import rpy2.robjects as robjects
import numpy as np
from rpy2.robjects import pandas2ri
pandas2ri.activate()

# load your file
robjects.r['load']('Data.RDa')

matrix = robjects.r['data']

matrix

I get the following results:

R object with classes: ('data.frame',) mapped to:
<DataFrame - Python:0x0CF46F58 / R:0x0ED0F200>
[Float..., Float..., Float..., ..., Float..., Float..., Float...]
  area: <class 'rpy2.robjects.vectors.FloatVector'>
  R object with classes: ('numeric',) mapped to:
<FloatVector - Python:0x0CF56A80 / R:0x0F281898>
[NA_real_, NA_real_, NA_real_, ..., NA_real_, NA_real_, NA_real_]
  i: <class 'rpy2.robjects.vectors.FloatVector'>
  R object with classes: ('numeric',) mapped to:
<FloatVector - Python:0x0CF68E68 / R:0x0F2B9520>
[NA_real_, NA_real_, NA_real_, ..., NA_real_, NA_real_, NA_real_]
  s: <class 'rpy2.robjects.vectors.FloatVector'>
  R object with classes: ('numeric',) mapped to:
<FloatVector - Python:0x0CF68940 / R:0x0F380008>
[NA_real_, NA_real_, NA_real_, ..., NA_real_, NA_real_, NA_real_]
  ...
  upslope_area: <class 'rpy2.robjects.vectors.FloatVector'>
  R object with classes: ('numeric',) mapped to:
<FloatVector - Python:0x0D03FDA0 / R:0x0FE87C90>
[NA_real_, NA_real_, NA_real_, ..., 292.256494, NA_real_, NA_real_]
  i: <class 'rpy2.robjects.vectors.FloatVector'>
  R object with classes: ('numeric',) mapped to:
<FloatVector - Python:0x0D03FC88 / R:0x0FEBF918>
[331347.500000, 331352.500000, 331357.500000, ..., 332187.500000, 332192.500000, 332197.500000]
  s: <class 'rpy2.robjects.vectors.FloatVector'>
  R object with classes: ('numeric',) mapped to:
<FloatVector - Python:0x0D03FE68 / R:0x0FEF75A0>
[4554812.500000, 4554812.500000, 4554812.500000, ..., 4553982.500000, 4553982.500000, 4553982.500000]

How do I convert this to a pandas data frame?

Answer 1

This looks like a missing call to the current conversion when retrieving from the search path the first R object with the symbol 'data' (in short, when doing robjects.r["data"] ). Open an issue on the rpy2 tracker if there isn't already one, or make noise in the comments for the issue already opened if unresolved or assumed to be resolved prematurely.

Calling explicitly conversion rules limited to a code block should make an easy workaround, and may be help you ensure good performances. The conversion mechanism provides convenience, but often at the expense of performances as a copy of the data frame is made each time in either direction the conversion is going.

Here is what would look like:

from rpy2.robjects import default_converter
from rpy2.robjects import pandas2ri
from rpy2.robjects.conversion import localconverter

# use the default conversion rules to which the pandas conversion
# is added
with localconverter(default_converter + pandas2ri.converter) as cv:
    dataf = robjects.r["data"]

This is in the doc: http://rpy2.readthedocs.io/en/version_2.8.x/robjects_convert.html#local-conversion-rules