Validation Parameters

openenergyid/mvlr/__init__.py

@@ -1,11 +1,12 @@
 """Multi-variable linear regression (MVLR) module."""
 
-from .mvlr import MultiVariableLinearRegression, find_best_mvlr
+from .mvlr import MultiVariableLinearRegression, find_best_mvlr, ValidationParameters
 from .models import IndependentVariable, MultiVariableRegressionResult
 
 __all__ = [
     "MultiVariableLinearRegression",
     "MultiVariableRegressionResult",
     "IndependentVariable",
     "find_best_mvlr",
+    "ValidationParameters",
 ]

openenergyid/mvlr/models.py

@@ -10,6 +10,14 @@
 from .mvlr import MultiVariableLinearRegression
 
 
+class ConfidenceInterval(BaseModel):
+    """Confidence interval for a coefficient."""
+
+    confidence: float
+    lower: float
+    upper: float
+
+
 class IndependentVariable(BaseModel):
     """Independent variable for a multivariable linear regression model."""
 
@@ -18,7 +26,7 @@ class IndependentVariable(BaseModel):
     t_stat: Optional[float] = None
     p_value: Optional[float] = None
     std_err: Optional[float] = None
-    confidence_interval: Optional[dict[str, float]] = None
+    confidence_interval: Optional[ConfidenceInterval] = None
 
     @classmethod
     def from_fit(cls, fit: fm.ols, name: str) -> "IndependentVariable":
@@ -29,11 +37,11 @@ def from_fit(cls, fit: fm.ols, name: str) -> "IndependentVariable":
             t_stat=fit.tvalues[name],
             p_value=fit.pvalues[name],
             std_err=fit.bse[name],
-            confidence_interval={
-                "confidence": 0.95,
-                "lower": fit.conf_int().transpose()[name][0],
-                "upper": fit.conf_int().transpose()[name][1],
-            },
+            confidence_interval=ConfidenceInterval(
+                confidence=0.95,
+                lower=fit.conf_int().transpose()[name][0],
+                upper=fit.conf_int().transpose()[name][1],
+            ),
         )
 
 

openenergyid/mvlr/mvlr.py

@@ -3,6 +3,7 @@
 
 import numpy as np
 import pandas as pd
+from pydantic import BaseModel, Field
 import statsmodels.formula.api as fm
 from patsy import LookupFactor, ModelDesc, Term  # pylint: disable=no-name-in-module
 from statsmodels.sandbox.regression.predstd import wls_prediction_std
@@ -12,6 +13,20 @@
 from .helpers import resample_input_data
 
 
+class ValidationParameters(BaseModel):
+    """Parameters for validation of a multivariable linear regression model."""
+
+    rsquared: float = Field(
+        0.75, ge=0, le=1, description="Minimum acceptable value for the adjusted R-squared"
+    )
+    f_pvalue: float = Field(
+        0.05, ge=0, le=1, description="Maximum acceptable value for the F-statistic"
+    )
+    pvalues: float = Field(
+        0.05, ge=0, le=1, description="Maximum acceptable value for the p-values of the t-statistic"
+    )
+
+
 class MultiVariableLinearRegression:
     """Multi-variable linear regression.
 
@@ -41,7 +56,7 @@ def __init__(
         confint: float = 0.95,
         cross_validation: bool = False,
         allow_negative_predictions: bool = False,
-        validation_params: dict = None,
+        validation_params: ValidationParameters = None,
         granularity: Granularity = None,
     ):
         """Parameters
@@ -65,15 +80,8 @@ def __init__(
             If True, allow predictions to be negative.
             For gas consumption or PV production, this is not physical
             so allow_negative_predictions should be False
-        validation_params : dict, default=None
-            Dictionary with parameters to validate the model.
-            The following parameters are supported:
-            - "rsquared": float, default=0.75
-                Minimum acceptable value for the adjusted R-squared
-            - "f_pvalue": float, default=0.05
-                Maximum acceptable value for the F-statistic
-            - "pvalues": float, default=0.05
-                Maximum acceptable value for the p-values of the t-statistic
+        validation_params : ValidationParameters, default=None
+            Parameters to validate the model.
         """
         self.data = data.copy()
         if y not in self.data.columns:
@@ -87,7 +95,7 @@ def __init__(
         self.confint = confint
         self.cross_validation = cross_validation
         self.allow_negative_predictions = allow_negative_predictions
-        self.validation_params = validation_params or {}
+        self.validation_params = validation_params or ValidationParameters()
         self.granularity = granularity
         self._fit = None
         self._list_of_fits = []
@@ -400,16 +408,16 @@ def is_valid(self) -> bool:
         -------
             bool: True if the model is valid, False otherwise.
         """
-        if self.fit.rsquared_adj < self.validation_params.get("rsquared", 0.75):
+        if self.fit.rsquared_adj < self.validation_params.rsquared:
             return False
 
-        if self.fit.f_pvalue > self.validation_params.get("f_pvalue", 0.05):
+        if self.fit.f_pvalue > self.validation_params.f_pvalue:
             return False
 
         param_keys = self.fit.pvalues.keys().tolist()
         param_keys.remove("Intercept")
         for k in param_keys:
-            if self.fit.pvalues[k] > self.validation_params.get("pvalues", 0.05):
+            if self.fit.pvalues[k] > self.validation_params.pvalues:
                 return False
 
         return True