compare_featurestores.py

from featurestore import open_featurestore
from featurestore.domain.featurestores import FeaturestoreReference

# https://flyte.labs.pachama.com/console/projects/baseline/domains/development/executions/fc1hfynprlwrkk-n5-0?duration=all
dev_featurestore = open_featurestore(
    FeaturestoreReference(
        collection="gs://labs-featurestore/managed/9a873487-91e5-4e83-a965-f737de4ec2c6",
        group="group",
    )
)

# https://flyte.labs.pachama.com/console/projects/baseline/domains/staging/executions/f6pbmc135vcf6a-n5-0?duration=all
stage_featurestore = open_featurestore(
    FeaturestoreReference(
        collection="gs://labs-featurestore/managed/7d495458-6e14-414a-a58b-d7f6c6c8ac5e",
        group="group",
    )
)
for data_var in [
    "EnvironmentalFeatures_GPP",
    "ModisTreeCoverFeatures_Percent_Tree_Cover",
    "SRTMFeatures_slope",
    "StaticGFWFeatures_lossyear",
    "distance_to_StaticGFWFeatures_lossyear",
    "iucn_mask",
    "verra_carbon_project",
]:
    dev_var = dev_featurestore[data_var]
    dev_da = dev_var.data.compute()
    stage_var = stage_featurestore[data_var]
    stage_da = stage_var.data.compute()
    diff = (dev_da!=stage_da).sum()
    print(f"{data_var}: {dev_var.equals(stage_var)}, {diff/dev_da.size:.2%} diff")