3.1. Automate model's running and save results in csv file

Fernando-Urbano / monash-time-series-replication

Monash Time-Series Forecast in Reproducible Analytical Pipeline (RAP) format

0 stars 2 forks source link

Closed Fernando-Urbano closed 3 months ago

Fernando-Urbano commented 4 months ago

issue regarding base_path: not working (problem in running R).
had to change to python 3.11.
half of the models are in python.
issue with numpy.
can make the calculation of the error in the in python (once you have datapoints of the errors).

Fernando-Urbano commented 4 months ago

it was rightly pointed by Aben that new changes might have been necessary for python 3.8: gluonts has multiple dependencies that require an older version of python
python 3.7, as used by the researchers, is not available in conda

Fernando-Urbano commented 4 months ago

Aben assure that the configurations were good to run all R models: accuracy of results have been getting better as we adjust the files.
Aben fixed catboost and did results for the some databases
Fernando created script that allows user to decide for which models/databases he/she wants to run