Error while running PYAF

[dineshgit]

I'm planning to use PYAF for one of my projects and I am urgently trying to get this up and running, please help.

Upon running the command below:

import pyaf.ForecastEngine as autof lEngine = autof.cForecastEngine() lEngine.train(ozone_dataframe , 'Month' , 'Ozone', 12);

I got this error:

Could someone please help me with this error ?

[antoinecarme]

Hi @dineshgit ,

The code snippet you pasted does not work... name 'ozone_dataframe' is not defined

Otherwise, I am sorry, I cannot help with numpy dll errors. Google is your friend.

Good luck.

[dineshgit]

The code snippet is from the tutorial here:

https://github.com/antoinecarme/pyaf/blob/master/docs/PyAF_Introduction.ipynb

[antoinecarme]

Yes. I see, but the code in a jupyter notebook must be extracted/pasted starting from the beginning of the page !!!

[dineshgit]

import pandas as pd csvfile_link = "https://raw.githubusercontent.com/antoinecarme/TimeSeriesData/master/ozone-la.csv" ozone_dataframe = pd.read_csv(csvfile_link);

import datetime ozone_dataframe['Month'] = ozone_dataframe['Month'].apply(lambda x : datetime.datetime.strptime(x, "%Y-%m")) ozone_dataframe.head()

%matplotlib inline
ozone_dataframe.plot.line('Month', ['Ozone'], grid = True, figsize=(12, 8))

import pyaf.ForecastEngine as autof lEngine = autof.cForecastEngine() lEngine.train(ozone_dataframe , 'Month' , 'Ozone', 12);

At the last 3 lines, I'm getting the error I reported.

[antoinecarme]

Thank you for the full copy-paste ... appreciated ;)

However, this seems to be a windows dll install issue with numpy and anaconda ... cannot help. Sorry, this is one reason why I am a very motivated debian user !!!

[dineshgit]

The problem is that the locally installed MKL package is not compatible with numpy. The solution is to install numpy using conda install numpy instead of pip install numpy. After that, please proceed to install all the other libraries required for pyAF as documented in the installation sections.

PIP installation works well for everything, but to install Numpy use Conda Install.

To find out the MKL library numpy is linked against, run the below Python script in the activated conda environment.

import numpy as np
np.show_config()