68 lines
2.6 KiB
Python
68 lines
2.6 KiB
Python
# ================================================================
|
|
# services/forecasting/app/models/predictions.py
|
|
# ================================================================
|
|
"""
|
|
Additional prediction models for the forecasting service
|
|
"""
|
|
|
|
from sqlalchemy import Column, String, Integer, Float, DateTime, Boolean, Text, JSON
|
|
from sqlalchemy.dialects.postgresql import UUID
|
|
from datetime import datetime, timezone
|
|
import uuid
|
|
|
|
from shared.database.base import Base
|
|
|
|
class ModelPerformanceMetric(Base):
|
|
"""Track model performance over time"""
|
|
__tablename__ = "model_performance_metrics"
|
|
|
|
id = Column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
|
|
model_id = Column(UUID(as_uuid=True), nullable=False, index=True)
|
|
tenant_id = Column(UUID(as_uuid=True), nullable=False, index=True)
|
|
product_name = Column(String(255), nullable=False)
|
|
|
|
# Performance metrics
|
|
mae = Column(Float) # Mean Absolute Error
|
|
mape = Column(Float) # Mean Absolute Percentage Error
|
|
rmse = Column(Float) # Root Mean Square Error
|
|
accuracy_score = Column(Float)
|
|
|
|
# Evaluation period
|
|
evaluation_date = Column(DateTime(timezone=True), nullable=False)
|
|
evaluation_period_start = Column(DateTime(timezone=True))
|
|
evaluation_period_end = Column(DateTime(timezone=True))
|
|
|
|
# Metadata
|
|
sample_size = Column(Integer)
|
|
created_at = Column(DateTime(timezone=True), default=lambda: datetime.now(timezone.utc))
|
|
|
|
def __repr__(self):
|
|
return f"<ModelPerformanceMetric(model_id={self.model_id}, mae={self.mae})>"
|
|
|
|
class PredictionCache(Base):
|
|
"""Cache frequently requested predictions"""
|
|
__tablename__ = "prediction_cache"
|
|
|
|
id = Column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
|
|
cache_key = Column(String(255), unique=True, nullable=False, index=True)
|
|
|
|
# Cached data
|
|
tenant_id = Column(UUID(as_uuid=True), nullable=False, index=True)
|
|
product_name = Column(String(255), nullable=False)
|
|
location = Column(String(255), nullable=False)
|
|
forecast_date = Column(DateTime(timezone=True), nullable=False)
|
|
|
|
# Cached results
|
|
predicted_demand = Column(Float, nullable=False)
|
|
confidence_lower = Column(Float, nullable=False)
|
|
confidence_upper = Column(Float, nullable=False)
|
|
model_id = Column(UUID(as_uuid=True), nullable=False)
|
|
|
|
# Cache metadata
|
|
created_at = Column(DateTime(timezone=True), default=lambda: datetime.now(timezone.utc))
|
|
expires_at = Column(DateTime(timezone=True), nullable=False)
|
|
hit_count = Column(Integer, default=0)
|
|
|
|
def __repr__(self):
|
|
return f"<PredictionCache(key={self.cache_key}, product={self.product_name})>"
|