junte/Model
1
0
Fork 0
mirror of https://github.com/Hestia-Homes/Model.git synced 2026-06-08 11:17:27 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Updating router for chunked scoring

Browse source
This commit is contained in:
Khalim Conn-Kowlessar 2024-03-28 15:54:31 +00:00
parent 22a3e21f52
commit 8dbd69eef9
2 changed files with 27 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
backend/Property.py
View file

@ -233,7 +233,7 @@ class Property:
output["walls_thermal_transmittance_ending"] = recommendation["new_u_value"]
# Setting the insulation thickness here to above average should be tested further because we
# don't see a high volume of instances for this
output["walls_insulation_thickness_ending"] = "above average"
output["walls_insulation_thickness_ending"] = "average"
output["walls_energy_eff_ending"] = "Good"
# Note: often when the wall is insulatied, the internal/external insulation is not noted so we should

36
backend/app/plan/router.py
View file

@ -41,6 +41,7 @@ from backend.ml_models.Valuation import PropertyValuation
logger = setup_logger()
BATCH_SIZE = 5
SCORING_BATCH_SIZE = 400
def patch_epc(config, epc_records):
@ -164,7 +165,7 @@ async def trigger_plan(body: PlanTriggerRequest):
recommendations = {}
recommendations_scoring_data = []
representative_recommendations = {}
for p in input_properties:
for p in tqdm(input_properties):
# Property recommendations
p.get_components(cleaned, photo_supply_lookup, floor_area_decile_thresholds)
@ -196,15 +197,30 @@ async def trigger_plan(body: PlanTriggerRequest):
model_api = ModelApi(portfolio_id=body.portfolio_id, timestamp=created_at)
all_predictions = model_api.predict_all(
df=recommendations_scoring_data,
bucket=get_settings().DATA_BUCKET,
prediction_buckets={
"sap_change_predictions": get_settings().SAP_PREDICTIONS_BUCKET,
"heat_demand_predictions": get_settings().HEAT_PREDICTIONS_BUCKET,
"carbon_change_predictions": get_settings().CARBON_PREDICTIONS_BUCKET
}
)
all_predictions = {
"sap_change_predictions": pd.DataFrame(),
"heat_demand_predictions": pd.DataFrame(),
"carbon_change_predictions": pd.DataFrame()
}
to_loop_over = range(0, recommendations_scoring_data.shape[0], SCORING_BATCH_SIZE)
for chunk in tqdm(to_loop_over, total=len(to_loop_over)):
predictions_dict = model_api.predict_all(
df=recommendations_scoring_data.iloc[chunk:chunk + SCORING_BATCH_SIZE],
bucket=get_settings().DATA_BUCKET,
prediction_buckets={
"sap_change_predictions": get_settings().SAP_PREDICTIONS_BUCKET,
"heat_demand_predictions": get_settings().HEAT_PREDICTIONS_BUCKET,
"carbon_change_predictions": get_settings().CARBON_PREDICTIONS_BUCKET
}
)
# Append the predictions to the predictions dictionary
for key, scored in predictions_dict.items():
all_predictions[key] = pd.concat([all_predictions[key], scored])
# TODO: TEMP
# all_predictions["heat_demand_predictions"] = all_predictions["sap_change_predictions"].copy()
# all_predictions["carbon_change_predictions"] = all_predictions["sap_change_predictions"].copy()
# Insert the predictions into the recommendations and run the optimiser
# TODO: If a recommendation has a negative impact on SAP, we should remove it - this seems to have become a