|
|
"""Gateway notebook for https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting""" |
|
|
|
|
|
import os |
|
|
|
|
|
import kaggle_evaluation.core.base_gateway |
|
|
import kaggle_evaluation.core.templates |
|
|
import polars as pl |
|
|
|
|
|
|
|
|
class JSGateway(kaggle_evaluation.core.templates.Gateway): |
|
|
def __init__(self, data_paths: tuple[str, str] | None = None): |
|
|
super().__init__(data_paths, file_share_dir=None) |
|
|
self.data_paths = data_paths |
|
|
self.set_response_timeout_seconds(60) |
|
|
|
|
|
def unpack_data_paths(self): |
|
|
if not self.data_paths: |
|
|
self.test_path = ( |
|
|
"/kaggle/input/jane-street-realtime-marketdata-forecasting/test.parquet" |
|
|
) |
|
|
self.lags_path = ( |
|
|
"/kaggle/input/jane-street-realtime-marketdata-forecasting/lags.parquet" |
|
|
) |
|
|
else: |
|
|
self.test_path, self.lags_path = self.data_paths |
|
|
|
|
|
def generate_data_batches(self): |
|
|
date_ids = sorted( |
|
|
pl.scan_parquet(self.test_path) |
|
|
.select(pl.col("date_id").unique()) |
|
|
.collect() |
|
|
.get_column("date_id") |
|
|
) |
|
|
assert date_ids[0] == 0 |
|
|
|
|
|
for date_id in date_ids: |
|
|
test_batches = pl.read_parquet( |
|
|
os.path.join(self.test_path, f"date_id={date_id}"), |
|
|
).group_by("time_id", maintain_order=True) |
|
|
|
|
|
lags = pl.read_parquet( |
|
|
os.path.join(self.lags_path, f"date_id={date_id}"), |
|
|
) |
|
|
|
|
|
for (time_id,), test in test_batches: |
|
|
test_data = (test, lags if time_id == 0 else None) |
|
|
validation_data = test.select('row_id') |
|
|
yield test_data, validation_data |
|
|
|
|
|
|
|
|
if __name__ == "__main__": |
|
|
if os.getenv("KAGGLE_IS_COMPETITION_RERUN"): |
|
|
gateway = JSGateway() |
|
|
|
|
|
gateway.run() |
|
|
else: |
|
|
print("Skipping run for now") |
|
|
|
