Merge pull request #636 from henrydingliu/master

Adding a test for drop_valuation

Author: henrydingliu

chainladder/development/barnzehn.py

@@ -75,7 +75,7 @@ def fit(self, X, y=None, sample_weight=None):
         self.model_ = DevelopmentML(Pipeline(steps=[
             ('design_matrix', PatsyFormula(self.formula)),
             ('model', LinearRegression(fit_intercept=False))]),
-                    y_ml=response, fit_incrementals=False, feat_eng = self.feat_eng, drop=self.drop, drop_valuation = self.drop_valuation, weighted_step = 'model').fit(tri)
+                    y_ml=response, fit_incrementals=True, feat_eng = self.feat_eng, drop=self.drop, drop_valuation = self.drop_valuation, weighted_step = 'model').fit(X = tri, sample_weight = sample_weight)
         resid = tri - self.model_.triangle_ml_[
             self.model_.triangle_ml_.valuation <= tri.valuation_date]
         self.mse_resid_ = (resid**2).sum(0).sum(1).sum(2).sum() / (
@@ -100,7 +100,7 @@ def transform(self, X):
             X_new : New triangle with transformed attributes.
         """
         X_new = X.copy()
-        X_ml, weight_ml = self.model_._prep_X_ml(X.cum_to_incr().log())
+        X_ml = self.model_._prep_X_ml(X.cum_to_incr().log())
         y_ml = self.model_.estimator_ml.predict(X_ml)
         triangle_ml, predicted_data = self.model_._get_triangle_ml(X_ml, y_ml)
         backend = "cupy" if X.array_backend == "cupy" else "numpy"

chainladder/development/glm.py

@@ -22,15 +22,16 @@ class TweedieGLM(DevelopmentBase):
 
     Parameters
     ----------
+    drop: tuple or list of tuples
+        Drops specific origin/development combination(s)
+    drop_valuation: str or list of str (default = None)
+        Drops specific valuation periods. str must be date convertible.
     design_matrix: formula-like
         A patsy formula describing the independent variables, X of the GLM
     response:  str
         Column name for the reponse variable of the GLM.  If ommitted, then the
         first column of the Triangle will be used.
-    weight: str
-        Column name of any weight to use in the GLM. If none specified, then an
-        unweighted regression will be performed.
-    power: float, default=0
+    power: float, default=1
             The power determines the underlying target distribution according
             to the following table:
             +-------+------------------------+
@@ -52,7 +53,7 @@ class TweedieGLM(DevelopmentBase):
         regularization strength. ``alpha = 0`` is equivalent to unpenalized
         GLMs. In this case, the design matrix `X` must have full column rank
         (no collinearities).
-    link: {'auto', 'identity', 'log'}, default='auto'
+    link: {'auto', 'identity', 'log'}, default='log'
         The link function of the GLM, i.e. mapping from linear predictor
         `X @ coeff + intercept` to prediction `y_pred`. Option 'auto' sets
         the link depending on the chosen family as follows:
@@ -78,10 +79,11 @@ class TweedieGLM(DevelopmentBase):
     """
 
     def __init__(self, design_matrix='C(development) + C(origin)',
-                 response=None, weight=None, power=1.0, alpha=1.0, link='log',
-                 max_iter=100, tol=0.0001, warm_start=False, verbose=0):
+                 response=None, power=1.0, alpha=1.0, link='log',
+                 max_iter=100, tol=0.0001, warm_start=False, verbose=0, drop=None,drop_valuation=None):
+        self.drop = drop
+        self.drop_valuation = drop_valuation
         self.response=response
-        self.weight=weight
         self.design_matrix = design_matrix
         self.power=power
         self.alpha=alpha
@@ -93,13 +95,18 @@ def __init__(self, design_matrix='C(development) + C(origin)',
 
     def fit(self, X, y=None, sample_weight=None):
         response = X.columns[0] if not self.response else self.response
+        if sample_weight is None:
+            weight = None
+        else:
+            weight = 'model'
         self.model_ = DevelopmentML(Pipeline(steps=[
             ('design_matrix', PatsyFormula(self.design_matrix)),
             ('model', TweedieRegressor(
                     link=self.link, power=self.power, max_iter=self.max_iter,
                     tol=self.tol, warm_start=self.warm_start,
                     verbose=self.verbose, fit_intercept=False))]),
-                    y_ml=response, weight_ml=self.weight).fit(X)
+                    y_ml=response, weighted_step = weight,
+                    drop=self.drop, drop_valuation=self.drop_valuation).fit(X = X, sample_weight = sample_weight)
         return self
 
     @property

chainladder/development/learning.py

@@ -33,6 +33,8 @@ class DevelopmentML(DevelopmentBase):
         Time Series aspects of the model. Predictions from one development period
         get used as featues in the next development period. Lags should be negative
         integers.
+    weight_step: str
+        Step name within estimator_ml that is weighted
     drop: tuple or list of tuples
         Drops specific origin/development combination(s)
     drop_valuation: str or list of str (default = None)
@@ -56,8 +58,7 @@ def test_func(df)
             return df['origin'] + 1
         )
     fit_incrementals:
-        Whether the response variable should be converted to an incremental basis
-        for fitting.
+        Whether the response variable should be converted to an incremental basis for fitting.
 
     Attributes
     ----------
@@ -70,10 +71,9 @@ def test_func(df)
     """
 
     def __init__(self, estimator_ml=None, y_ml=None, autoregressive=False,
-                 weight_ml=None, weighted_step=None,drop=None,drop_valuation=None,fit_incrementals=True, feat_eng=None):
+                 weighted_step=None,drop=None,drop_valuation=None,fit_incrementals=True, feat_eng=None):
         self.estimator_ml=estimator_ml
         self.y_ml=y_ml
-        self.weight_ml = weight_ml
         self.weighted_step = weighted_step
         self.autoregressive = autoregressive
         self.drop = drop
@@ -168,7 +168,7 @@ def _prep_X_ml(self, X):
         df_base = X.incr_to_cum().to_frame(
             keepdims=True, implicit_axis=True, origin_as_datetime=True
             ).reset_index().iloc[:, :-1]
-        df = df_base.merge(X.cum_to_incr().to_frame(
+        df = df_base.merge(X_.to_frame(
                 keepdims=True, implicit_axis=True, origin_as_datetime=True
             ).reset_index(), how='left',
             on=list(df_base.columns)).fillna(0)
@@ -177,13 +177,18 @@ def _prep_X_ml(self, X):
         if self.feat_eng is not None:            
             for key, item in self.feat_eng.items():
                 df[key] = item['func'](df=df,**item['kwargs'])
+        return df
+
+    def _prep_w_ml(self,X,sample_weight=None):
         weight_base = (~np.isnan(X.values)).astype(float)
-        weight = weight_base.copy()
+        weight = weight_base.copy()               
         if self.drop is not None:
             weight = weight * self._drop_func(X)
         if self.drop_valuation is not None:
-            weight = weight * self._drop_valuation_func(X)        
-        return df, weight.flatten()[weight_base.flatten()>0]
+            weight = weight * self._drop_valuation_func(X)
+        if sample_weight is not None:
+            weight = weight * sample_weight.values 
+        return weight.flatten()[weight_base.flatten()>0]
 
     def fit(self, X, y=None, sample_weight=None):
         """Fit the model with X.
@@ -194,8 +199,8 @@ def fit(self, X, y=None, sample_weight=None):
             Set of LDFs to which the estimator will be applied.
         y : None
             Ignored, use y_ml to set a reponse variable for the ML algorithm
-        sample_weight : None
-            Ignored
+        sample_weight : Triangle-like
+            Weights to use in the regression
 
         Returns
         -------
@@ -214,8 +219,9 @@ def fit(self, X, y=None, sample_weight=None):
         self.valuation_encoder_ = dict(zip(
             val,
             (pd.Series(val).rank()-1)/{'Y':1, 'S': 2, 'Q':4, 'M': 12}[X.development_grain]))
-        df, weight = self._prep_X_ml(X)
+        df = self._prep_X_ml(X)
         self.df_ = df
+        weight = self._prep_w_ml(X,sample_weight)
         self.weight_ = weight
         if self.weighted_step == None:
             sample_weights = {}
@@ -249,7 +255,7 @@ def transform(self, X):
             X_new : New triangle with transformed attributes.
         """
         X_new = X.copy()
-        X_ml, weight_ml = self._prep_X_ml(X)
+        X_ml = self._prep_X_ml(X)
         y_ml=self.estimator_ml.predict(X_ml)
         triangle_ml, predicted_data = self._get_triangle_ml(X_ml, y_ml)
         backend = "cupy" if X.array_backend == "cupy" else "numpy"

chainladder/development/tests/test_barnzehn.py

@@ -1,9 +1,9 @@
 import numpy as np
 import chainladder as cl
 import pytest
+abc = cl.load_sample('abc')
 
 def test_basic_bz():
-    abc = cl.load_sample('abc')
     assert np.all(
         np.around(cl.BarnettZehnwirth(formula='C(origin)+C(development)').fit(abc).coef_.T.values,3).flatten()
         == np.array([11.837,0.179,0.345,0.378,0.405,0.427,0.431,0.66,0.963,1.157,1.278,0.251,-0.056,-0.449,-0.829,-1.169,-1.508,-1.798,-2.023,-2.238,-2.428])
@@ -21,7 +21,6 @@ def test_feat_eng_1():
     def test_func(df):
         return df["development"]
 
-    abc = cl.load_sample('abc')
     test_dict = {'testfeat':{'func':test_func,'kwargs':{}}}
 
     assert np.all(
@@ -38,18 +37,30 @@ def test_feat_eng_2():
     def origin_onehot(df,ori):
         return [1 if x == ori else 0 for x in df["origin"]]
 
-    abc = cl.load_sample('abc')
     feat_dict = {f'origin_{x}':{'func':origin_onehot,'kwargs':{'ori':float(x+1)}} for x in range(10)}
     assert np.all(
         np.around(cl.BarnettZehnwirth(formula='+'.join([f'C({x})' for x in feat_dict.keys()]),feat_eng = feat_dict).fit(abc).ldf_.values,3)
         == np.around(cl.BarnettZehnwirth(formula='C(origin)').fit_transform(abc).ldf_.values,3)
     )
 
+def test_drops():
+    '''
+    this function tests the passing in a basic drop_valuation
+    '''
+    def test_func(df):
+        return df["development"]
+
+    test_dict = {'testfeat':{'func':test_func,'kwargs':{}}}
+
+    assert np.all(
+        np.around(cl.BarnettZehnwirth(formula='C(development)',drop_valuation='1979').fit(abc).triangle_ml_.values,3)
+        == np.around(cl.BarnettZehnwirth(formula='C(testfeat)',drop = [('1977',36),('1978',24),('1979',12)],feat_eng = test_dict).fit(abc).triangle_ml_.values,3)
+    )
+
 def test_bz_2008():
     '''
     this function tests the drop parameter by recreating the example in the 2008 BZ paper, section 4.1
     '''
-    abc = cl.load_sample('abc')
     exposure=np.array([[2.2], [2.4], [2.2], [2.0], [1.9], [1.6], [1.6], [1.8], [2.2], [2.5], [2.6]])
     abc_adj = abc/exposure
 

chainladder/development/tests/test_glm.py

@@ -5,3 +5,9 @@ def test_basic_odp_cl(genins):
         (cl.Chainladder().fit(genins).ultimate_ -
          cl.Chainladder().fit(cl.TweedieGLM().fit_transform(genins)).ultimate_) /
         genins.latest_diagonal).max()< 1e-2
+    
+def test_sample_weight(genins):
+    assert abs(
+        (cl.Chainladder().fit(genins).ultimate_ -
+         cl.Chainladder().fit(cl.TweedieGLM().fit_transform(genins,sample_weight=genins/genins)).ultimate_) /
+        genins.latest_diagonal).max()< 1e-2

chainladder/development/tests/test_learning.py

@@ -0,0 +1,19 @@
+import chainladder as cl
+from sklearn.linear_model import LinearRegression
+from sklearn.pipeline import Pipeline
+from chainladder.utils.utility_functions import PatsyFormula
+
+def test_incremental(genins):
+    response = [genins.columns[0]]
+    model = cl.DevelopmentML(Pipeline(steps=[
+        ('design_matrix', PatsyFormula('C(development)')),
+        ('model', LinearRegression(fit_intercept=False))]),
+                y_ml=response,fit_incrementals=False).fit(genins)
+    assert abs(model.triangle_ml_.loc[:,:,'2010',:] - genins.mean()).max() < 1e2
+
+def test_misc(genins):
+    model = cl.DevelopmentML(Pipeline(steps=[
+        ('design_matrix', PatsyFormula('C(development)')),
+        ('model', LinearRegression(fit_intercept=False))]),
+                weighted_step = ['model'], fit_incrementals=False).fit(genins, sample_weight=genins/genins)
+    assert abs(model.triangle_ml_.loc[:,:,'2010',:] - genins.mean()).max() < 1e2