How to utilize packages built in forecasting plugin
swapnilk
Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Registered Posts: 2 ✭✭✭
After using the forecast plugin, the models get stored in a folder with the name of model.pk.gz, unzipping it to model.pk but this binary file is inaccessible.
Answers
-
Alexandru Dataiker, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 1,225 Dataiker
Hi, Welcome to the Dataiku Community! The model.pk file is a pickle binary file.You can use it using python. Could please elaborate what exactly you are trying to do?
import pickle # import module first from gluonts.dataset.common import ListDataset f = open('gluon_train_dataset.pk', 'rb') # read mylist = pickle.load(f) # load file f.close() # This will be gluonts.dataset https://ts.gluon.ai/api/gluonts/gluonts.dataset.common.html
.
-
Hi Alex,
Thanks for the reply. I'm trying to access the codes written in model.pk file generated by forecast plugin.
-
Alexandru Dataiker, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 1,225 Dataiker
Hi,
The model.pk file in our case would be in my example is gluonts.model.trivial.identity.IdentityPredictor object. The code of the actual plugin is available here.
import pickle from gluonts.dataset import common f = open('model.pk', 'rb') # 'r' for reading; can be omitted pickle_data = pickle.load(f) # load file content as mydict print(pickle_data.__dict__ ) #example output {'__init_args__': OrderedDict([('freq', 'D'), ('num_samples', 100), ('prediction_length', 1)]), 'prediction_length': 1, 'freq': 'D', 'lead_time': 0, 'num_samples': 100}
Let me know if there is anything further I help with.