0

I have an R script which takes as input an excel file with two columns containing dates-values and it gives as output 3 dates with the corresponding prediction values. I have already successfully implemented it in Azure Machine Learning Studio using three nodes. One containing the zipped packages I use, one with the input .csv file and the last one with the R script. The problem is when I deploy it as a web service and I try to give as input new values for Col1 and Col2, I receive the following error.

FailedToParseValue: Failed to parse value '90000, 950000, 970000' as type 'System.Double'., Error code: LibraryExecutionError, Http status code:400

The zipped libraries I use attached are: Hmisc, gdata, forecast, lubridate, fma, expsmooth, ggplot2, tsibble, fpp2, and plyr. I have also tried using the notebooks provided but no good luck as I always face some kind of problem with package installation. Moreover, I tried to follow this approach https://azure.github.io/azureml-sdk-for-r/articles/train-and-deploy-to-aci/train-and-deploy-to-aci.html locally from R Studio but I have difficulty in adapting it to my case.
Any help would be greatly appreciated!

enter image description here enter image description here

XIpagII
  • 11
  • 2

1 Answers1

0

I didn't have any success installing packages via a zip. However the following worked for me in an Execute R Script and installed all dependancies also.

if(!require(package)) install.packages("package",repos = "https://ftp.heanet.ie/mirrors/cran.r-project.org/")

Make sure that your repo is from CRAN in your country.

VincentT
  • 1
  • 2