Merge pull request #85 from antoinedemathelin/master

docs: Updates doc and test

Author: antoinedemathelin

src_docs/_static/images/contribute.png

@@ -24,6 +24,7 @@
 <ul>
 <li class="toctree-l1"><a class="reference internal" href="{{ pathto("map") }}">Choosing the right algorithm</a></li>
 <li class="toctree-l1"><a class="reference internal" href="{{ pathto("examples/Quick_start") }}">Quick-Start</a></li>
+<li class="toctree-l1"><a class="reference internal" href="{{ pathto("examples/Developer_Guide") }}">Developer Guide</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">API Documentation</span></p>
 <ul>

src_docs/_static/images/fork.png

@@ -2,6 +2,7 @@
 Test functions for iwn module.
 """
 
+import numpy as np
 from sklearn.linear_model import RidgeClassifier
 from adapt.utils import make_classification_da
 from adapt.instance_based import IWC
@@ -15,6 +16,9 @@ def test_iwn():
     model.fit(Xs, ys);
     model.predict(Xt)
     model.score(Xt, yt)
+    w1 = model.predict_weights()
+    w2 = model.predict_weights(Xs)
+    assert np.abs(w1-w2).sum() < 10**-5
 
 
 def test_default_classif():
@@ -23,6 +27,9 @@ def test_default_classif():
     model.fit(Xs, ys);
     model.predict(Xt)
     model.score(Xt, yt)
+    w1 = model.predict_weights()
+    w2 = model.predict_weights(Xs)
+    assert np.abs(w1-w2).sum() < 10**-5
 
 
 def test_nn_classif():
@@ -32,3 +39,7 @@ def test_nn_classif():
     model.fit(Xs, ys);
     model.predict(Xt)
     model.score(Xt, yt)
+    w1 = model.predict_weights()
+    w2 = model.predict_weights(Xs)
+    assert np.abs(w1-w2).sum() < 10**-5
+    

src_docs/_templates/layout.html

@@ -5,6 +5,8 @@
 from sklearn.linear_model import RidgeClassifier
 from adapt.utils import make_classification_da
 from adapt.instance_based import IWN
+from adapt.utils import get_default_task
+from sklearn.neighbors import KNeighborsClassifier
 
 Xs, ys, Xt, yt = make_classification_da()
 
@@ -15,3 +17,21 @@ def test_iwn():
     model.score(Xt, yt)
     model.predict(Xs)
     model.predict_weights(Xs)
+    
+    
+def test_iwn_fit_estim():
+    task = get_default_task()
+    task.compile(optimizer="adam", loss="mse", metrics=["mae"])
+    model = IWN(task, Xt=Xt, sigma_init=0.1, random_state=0,
+                pretrain=True, pretrain__epochs=100, pretrain__verbose=0)
+    model.fit(Xs, ys)
+    model.score(Xt, yt)
+    model.predict(Xs)
+    model.predict_weights(Xs)
+    
+    model = IWN(KNeighborsClassifier(), Xt=Xt, sigma_init=0.1, random_state=0,
+                pretrain=True, pretrain__epochs=100, pretrain__verbose=0)
+    model.fit(Xs, ys)
+    model.score(Xt, yt)
+    model.predict(Xs)
+    model.predict_weights(Xs)

src_docs/examples/Developer_Guide.ipynb

@@ -3,6 +3,7 @@
 from sklearn.tree import DecisionTreeClassifier
 from sklearn.ensemble import RandomForestClassifier
 
+from adapt.utils import make_classification_da
 from adapt.parameter_based import TransferTreeClassifier, TransferForestClassifier
 
 methods = [
@@ -188,4 +189,16 @@ def test_transfer_tree():
         print('Testing score transferred model ({}) : {:.3f}'.format(method, score))
         clfs.append(transferred_dt.estimator)
         #clfs.append(clf_transfer)
-        scores.append(score)
+        scores.append(score)
+        
+        
+def test_specific():
+    Xs, ys, Xt, yt = make_classification_da()
+    
+    src_model = DecisionTreeClassifier()
+    src_model.fit(Xs, ys)
+    
+    model = TransferTreeClassifier(src_model)
+    
+    model.estimator_ = src_model
+    model._strut(Xt[:0], yt[:0], no_prune_on_cl=True, cl_no_prune=[0], node=3)

tests/test_iwc.py

@@ -11,11 +11,31 @@ def test_ulsif():
     model.fit(Xs, ys);
     model.predict(Xs)
     model.score(Xt, yt)
+    model.predict_weights()
+    model.predict_weights(Xs)
+    
+    model = ULSIF(RidgeClassifier(0.), Xt=Xt, kernel="rbf",
+                   lambdas=[0.1, 1., 10.], gamma=[0.1, 1., 10.], random_state=0)
+    model.fit(Xs, ys);
+    
+    model = ULSIF(RidgeClassifier(0.), Xt=Xt, kernel="rbf",
+                   lambdas=[0.1, 1., 10.], gamma=[0.1, 1., 10.], random_state=0)
+    model.fit(Xs[:73], ys[:73]);
 
 
 def test_rulsif():
     model = RULSIF(RidgeClassifier(0.), Xt=Xt, kernel="rbf", alpha=0.1,
                    lambdas=[0.1, 1., 10.], gamma=[0.1, 1., 10.], random_state=0)
     model.fit(Xs[:73], ys[:73]);
     model.predict(Xs)
-    model.score(Xt, yt)
+    model.score(Xt, yt)
+    model.predict_weights()
+    model.predict_weights(Xs)
+    
+    model = RULSIF(RidgeClassifier(0.), Xt=Xt, kernel="rbf", alpha=0.1,
+                   lambdas=[0.1, 1., 10.], gamma=[0.1, 1., 10.], random_state=0)
+    model.fit(Xs, ys);
+    
+    model = RULSIF(RidgeClassifier(0.), Xt=Xt[:73], kernel="rbf", alpha=0.1,
+                   lambdas=[0.1, 1., 10.], gamma=[0.1, 1., 10.], random_state=0)
+    model.fit(Xs, ys);