I had one more quick for the Optimization example....
# mlforecastb
Barrett Layman
01/04/2025, 1:05 PMI had one more quick for the Optimization example. Is it possible to use a distributed model like SparkLGBMForecast?
Barrett Layman
01/04/2025, 1:06 PMAt this stage, I received an error trying to pass SparkLGBMForecast().
Barrett Layman
01/04/2025, 1:06 PMTypeError: Cannot clone object 'SparkLGBMForecast_7d1c5816879c' (type <class 'mlforecast.distributed.models.spark.lgb.SparkLGBMForecast'>): it does not seem to be a scikit-learn estimator as it does not implement a 'get_params' method.
Barrett Layman
01/04/2025, 1:06 PMimport optuna
from mlforecast.optimization import mlforecast_objective
import mlforecast
from mlforecast.distributed.models.spark.lgb import SparkLGBMForecast
optuna.logging.set_verbosity(optuna.logging.WARNING)
objective = mlforecast_objective(
df=SparkDataFrame,
config_fn=config_fn,
loss=loss,
model=SparkLGBMForecast(),
freq='D',
n_windows=4,
h=7
)
study = optuna.create_study(
direction='minimize', sampler=optuna.samplers.TPESampler(seed=0)
)
study.optimize(objective, n_trials=2)
best_cfg = study.best_trial.user_attrs['config']
k
Krystian W.
01/04/2025, 4:34 PMI'd just implement your own objective with Mlforecast.cv method, since it's exactly how it's done in a package