Usage

Connect to the API and upload a dataset:

from modep_client.client import Client
from modep_client.datasets import Datasets

client = Client(os.environ['MODEP_API_KEY'])
datasets = Datasets(client)

df_train = pd.read_csv('https://jgoode.s3.amazonaws.com/titanic/train.csv')
df_test = pd.read_csv('https://jgoode.s3.amazonaws.com/titanic/test.csv')

train_dset = datasets.upload(df_train, 'titanic_train')
test_dset = datasets.upload(df_test, 'titanic_test')

Train all tabular models on the dataset for a max of 30 minutes each:

from modep_client.automl import FrameworkFlights

flights = FrameworkFlights(client)

frameworks = [
   'AutoGluon',
   'AutoGluon_bestquality',
   'autosklearn',
   'autosklearn2',
   'AutoWEKA',
   'constantpredictor',
   'DecisionTree',
   'flaml',
   'GAMA',
   'H2OAutoML',
   'hyperoptsklearn',
   'mljarsupervised',
   'mljarsupervised_compete',
   'MLNet',
   'RandomForest',
   'TPOT',
   'TunedRandomForest',
]

# starts the training
flight_job = flights.train(frameworks, train_dset['id'], test_dset['id'], 'survived', 60*30)

# waits for training to complete
flight = flight_job.result()