how to run auto.arima using rpy2 - rpy2

Question

I want to call R's auto.arima function from Python. I think i have not yet fully understood this interface. Can someone help me here - to send a time series obj to R, call forecast related functions and get back the results?
This is what I have done so far:
from rpy2.robjects import r
from rpy2.robjects import pandas2ri
#create a python time series
count = range(1, 51)
df['count'] = count
df['date'] = pd.date_range('2016-01-01', '2016-02-19')
df.set_index('date', inlace = True)
df.sort_index(inplace = True)
pandas2ri.activate()
r_timeseries = pandas2ri.py2ri(df)
r('fit <- auto.arima(r_timeseries)')
I think I have to import some R packages (like forecast). Not sure how to go about doing that in Python, properly pass the python time series object to R etc.
In [63]: r_ts = pandas2ri.py2ri(df)
In [64]: r_ts
Out[64]:
<DataFrame - Python:0x1126a93f8 / R:0x7ff7bfa51bc8>
[IntVector]
X0: <class 'rpy2.robjects.vectors.IntVector'>
<IntVector - Python:0x1126a96c8 / R:0x7ff7be1af1c0>
[ 1, 2, 3, ..., 48, 49, 50]
And, when I attempt to call forecast
In [83]: x = r('forecast(r_ts)')
/Library/Python/2.7/site-packages/rpy2/robjects/functions.py:106: UserWarning: Error in forecast(r_ts) : object 'r_ts' not found
res = super(Function, self).__call__(*new_args, **new_kwargs)
---------------------------------------------------------------------------
RRuntimeError Traceback (most recent call last)
<ipython-input-83-0765ffc30741> in <module>()
----> 1 x = r('forecast(r_ts)')
/Library/Python/2.7/site-packages/rpy2/robjects/__init__.pyc in __call__(self, string)
319 def __call__(self, string):
320 p = _rparse(text=StrSexpVector((string,)))
--> 321 res = self.eval(p)
322 return conversion.ri2py(res)
323
/Library/Python/2.7/site-packages/rpy2/robjects/functions.pyc in __call__(self, *args, **kwargs)
176 v = kwargs.pop(k)
177 kwargs[r_k] = v
--> 178 return super(SignatureTranslatedFunction, self).__call__(*args, **kwargs)
179
180 pattern_link = re.compile(r'\\link\{(.+?)\}')
/Library/Python/2.7/site-packages/rpy2/robjects/functions.pyc in __call__(self, *args, **kwargs)
104 for k, v in kwargs.items():
105 new_kwargs[k] = conversion.py2ri(v)
--> 106 res = super(Function, self).__call__(*new_args, **new_kwargs)
107 res = conversion.ri2ro(res)
108 return res
RRuntimeError: Error in forecast(r_ts) : object 'r_ts' not found
I tried the following as well:
In [99]: f = r('forecast.auto.arima(r_ts)')
---------------------------------------------------------------------------
RRuntimeError Traceback (most recent call last)
<ipython-input-99-1c4610d2740d> in <module>()
----> 1 f = r('forecast.auto.arima(r_ts)')
/Library/Python/2.7/site-packages/rpy2/robjects/__init__.pyc in __call__(self, string)
319 def __call__(self, string):
320 p = _rparse(text=StrSexpVector((string,)))
--> 321 res = self.eval(p)
322 return conversion.ri2py(res)
323
/Library/Python/2.7/site-packages/rpy2/robjects/functions.pyc in __call__(self, *args, **kwargs)
176 v = kwargs.pop(k)
177 kwargs[r_k] = v
--> 178 return super(SignatureTranslatedFunction, self).__call__(*args, **kwargs)
179
180 pattern_link = re.compile(r'\\link\{(.+?)\}')
/Library/Python/2.7/site-packages/rpy2/robjects/functions.pyc in __call__(self, *args, **kwargs)
104 for k, v in kwargs.items():
105 new_kwargs[k] = conversion.py2ri(v)
--> 106 res = super(Function, self).__call__(*new_args, **new_kwargs)
107 res = conversion.ri2ro(res)
108 return res
RRuntimeError: Error in eval(expr, envir, enclos) :
could not find function "forecast.auto.arima"

Answer 1

you could try what I do
import rpy2.robjects as ro
from rpy2.robjects import pandas2ri
pandas2ri.activate()
ro.r('library(forecast)')
rdf = pandas2ri.py2ri(df)
ro.globalenv['r_timeseries'] = rdf
pred = ro.r('as.data.frame(forecast(auto.arima(r_timeseries),h=5))')
this way, you can handle pred as a data frame like this
Point Forecast Lo 80 Hi 80 Lo 95 Hi 95
51 51 51 51 51 51
52 52 52 52 52 52
53 53 53 53 53 53
54 54 54 54 54 54
55 55 55 55 55 55

Answer 2

In the first attempt you are telling R to use a variable r_ts that it does not now much about (the name r_ts is defined in your Python namespace), and in the second attempt you are added to this a function name R does not know anything about. Both error message are precisely reporting this as a problem.
Your first attempt could be rewritten as:
x = r('forecast')(r_ts)