mirror of
https://github.com/webclinic017/drift.git
synced 2026-07-27 18:57:55 +00:00
b5ddee8dce
* feat(HPO): added `run_hpo` script * fix(Linter): ran * feat(HPO): removed any reference to sweep (superseeded by optuna) * fix(HPO): optimize for sharpe * fix(Config): removed glassnode data, save trials from hpo * feat(Labelling): added three-balanced method works again * fix(BetSizing): set the correct class labels * fix(HPO): powerset should return what's expected, added two new normalization methods * fix(Linter): ran * fix(DataLoader): sort the dataframe when fetching data * fix(Config): only take z-score of other assets
40 lines
1.4 KiB
Python
40 lines
1.4 KiB
Python
from config.types import RawConfig
|
|
from run_pipeline import run_pipeline
|
|
from config.presets import get_default_config
|
|
import optuna
|
|
from utils.helpers import powerset
|
|
|
|
config = get_default_config()
|
|
config.save_models = False
|
|
|
|
|
|
def train(trial):
|
|
search_space = {
|
|
"n_features_to_select": trial.suggest_int("n_features_to_select", 10, 200),
|
|
"forecasting_horizon": trial.suggest_int("forecasting_horizon", 3, 300),
|
|
"own_features": trial.suggest_categorical(
|
|
"own_features", powerset(["z_score", "level_2", "level_1"])
|
|
),
|
|
"other_features": trial.suggest_categorical(
|
|
"other_features", powerset(["z_score", "level_2", "level_1"])
|
|
),
|
|
"labeling": trial.suggest_categorical(
|
|
"labeling", ["two_class", "three_class_balanced", "three_class_imbalanced"]
|
|
),
|
|
"scaler": trial.suggest_categorical(
|
|
"scaler",
|
|
["normalize", "minmax", "standardize", "robust", "box-cox", "quantile"],
|
|
),
|
|
}
|
|
trial_config = RawConfig(**vars(config) | search_space)
|
|
outcome, _ = run_pipeline(
|
|
project_name="test", with_wandb=False, raw_config=trial_config
|
|
)
|
|
return 1 - outcome.get_output_stats()["sharpe"]
|
|
|
|
|
|
study = optuna.create_study()
|
|
study.optimize(train, n_trials=40)
|
|
print(study.best_params)
|
|
study.trials_dataframe().to_csv("output/trials.csv")
|