Merge pull request #64 from Techtonique/partial-fit

Partial fit

examples/custom_partial_fit.py

@@ -0,0 +1,55 @@
+import os 
+import nnetsauce as ns 
+import numpy as np 
+from sklearn.datasets import load_diabetes, load_iris, load_wine
+from sklearn.model_selection import train_test_split
+from sklearn.linear_model import SGDRegressor, SGDClassifier, Perceptron
+from sklearn.gaussian_process.kernels import Matern
+from time import time 
+
+print(f"\n ----- Running: {os.path.basename(__file__)}... ----- \n")
+
+print(f"\n\n ----- iris ----- \n\n")
+
+data = load_iris()
+
+X = data.data
+y= data.target
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = .2, random_state = 123)
+
+
+regr = ns.CustomClassifier(SGDClassifier(), n_clusters=0)
+start = time()
+regr.fit(X_train, y_train)
+print(f"Elapsed: {time() - start}s")
+preds = regr.predict(X_test)
+print(regr.score(X_test, y_test))
+regr.partial_fit(X_test[0, :], y_test[0])
+preds = regr.predict(X_test[1:, :])
+print(regr.score(X_test[1:, :], y_test[1:]))
+regr.partial_fit(X_test[1, :], y_test[1])
+preds = regr.predict(X_test[2:, :])
+print(regr.score(X_test[2:, :], y_test[2:]))
+
+
+print(f"\n\n ----- diabetes ----- \n\n")
+
+data = load_diabetes()
+X = data.data
+y= data.target
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = .2, random_state = 123)
+
+
+regr = ns.CustomRegressor(SGDRegressor(), n_clusters=0, n_hidden_features=10)
+start = time()
+regr.fit(X_train, y_train)
+print(f"Elapsed: {time() - start}s")
+preds = regr.predict(X_test)
+print(np.sqrt(np.mean((preds - y_test)**2)))
+regr.partial_fit(X_test[0, :], y_test[0])
+preds = regr.predict(X_test[1:, :])
+print(np.sqrt(np.mean((preds - y_test[1:])**2)))
+regr.partial_fit(X_test[1, :], y_test[1])
+preds = regr.predict(X_test[2:, :])
+print(np.sqrt(np.mean((preds - y_test[2:])**2)))
+

examples/lazy_custom_deep_classification.py

@@ -26,7 +26,7 @@
                                             "RandomForestRegressor"])
 
     start = time()
-    models, predictions = clf.fit(X_train, X_test, y_train, y_test)
+    models = clf.fit(X_train, X_test, y_train, y_test)
     print(f"\nElapsed: {time() - start} seconds\n")
 
     print(models)
@@ -44,7 +44,7 @@
                                 n_hidden_features=3)
 
     start = time()
-    models, predictions = clf.fit(X_train, X_test, y_train, y_test)
+    models = clf.fit(X_train, X_test, y_train, y_test)
     print(f"\nElapsed: {time() - start} seconds\n")
 
     print(models)

examples/lazy_custom_deep_regression.py

@@ -18,14 +18,14 @@
 verbose=0, ignore_warnings=True, estimators=["ExtraTreesRegressor", 
                                              "RandomForestRegressor", 
                                              "LassoLarsIC"])
-models, predictions = regr2.fit(X_train, X_test, y_train, y_test)
+models = regr2.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr2.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["DeepCustomRegressor(LassoLarsIC)"])
 
 regr = ns.LazyDeepRegressor(n_layers=3, n_hidden_features=2, 
 verbose=0, ignore_warnings=True)
-models, predictions = regr.fit(X_train, X_test, y_train, y_test)
+models = regr.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["DeepCustomRegressor(LassoLarsIC)"])
@@ -44,14 +44,14 @@
                                              "RandomForestRegressor", 
                                              "LassoLarsIC"], 
                                              preprocess=True)
-models, predictions = regr2.fit(X_train, X_test, y_train, y_test)
+models = regr2.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr2.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["DeepCustomRegressor(LassoLarsIC)"])
 
 regr = ns.LazyDeepRegressor(n_layers=3, n_hidden_features=2, 
 verbose=0, ignore_warnings=True, preprocess=True)
-models, predictions = regr.fit(X_train, X_test, y_train, y_test)
+models = regr.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["DeepCustomRegressor(LassoLarsIC)"])

examples/lazy_custom_regression.py

@@ -13,21 +13,21 @@
 regr = ns.LazyRegressor(verbose=0, ignore_warnings=True, custom_metric=None, 
                         preprocess=True, 
                         estimators=["RandomForestRegressor", "ExtraTreesRegressor"])
-models, predictions = regr.fit(X_train, X_test, y_train, y_test)
+models = regr.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["CustomRegressor(RandomForestRegressor)"])
 
 regr = ns.LazyRegressor(verbose=0, ignore_warnings=True, custom_metric=None, 
                         preprocess=True)
-models, predictions = regr.fit(X_train, X_test, y_train, y_test)
+models = regr.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["CustomRegressor(RandomForestRegressor)"])
 
 regr2 = ns.LazyRegressor(verbose=0, ignore_warnings=True, custom_metric=None, 
                         preprocess=False)
-models, predictions = regr2.fit(X_train, X_test, y_train, y_test)
+models = regr2.fit(X_train, X_test, y_train, y_test)
 model_dictionary = regr2.provide_models(X_train, X_test, y_train, y_test)
 print(models)
 print(model_dictionary["CustomRegressor(RandomForestRegressor)"])

examples/lazy_deepmts.py

@@ -41,7 +41,7 @@
                       lags = 20, n_hidden_features=7, n_clusters=2,
                       show_progress=False, preprocess=False, 
                       estimators=["ElasticNetCV", "RidgeCV"])
-modelss = regr_mts3.fit(df_train, df_test)
+models = regr_mts3.fit(df_train, df_test)
 model_dictionary = regr_mts3.provide_models(df_train, df_test)
 print(models)
 print(model_dictionary["DeepMTS(ElasticNetCV)"])
@@ -50,7 +50,7 @@
                           n_layers=3,
                       lags = 20, n_hidden_features=7, n_clusters=2,
                       show_progress=False, preprocess=False)
-modelss = regr_mts.fit(df_train, df_test)
+models = regr_mts.fit(df_train, df_test)
 model_dictionary = regr_mts.provide_models(df_train, df_test)
 print(models)
 print(model_dictionary["DeepMTS(LinearSVR)"])
@@ -59,7 +59,7 @@
                            n_layers=3,
                       lags = 20, n_hidden_features=7, n_clusters=2,
                       show_progress=False, preprocess=True)
-modelss = regr_mts2.fit(df_train, df_test)
+models = regr_mts2.fit(df_train, df_test)
 model_dictionary = regr_mts.provide_models(df_train, df_test)
 print(models)
 print(model_dictionary["DeepMTS(LinearSVR)"])
@@ -69,7 +69,7 @@
                       lags = 20, n_hidden_features=7, n_clusters=2,
                       replications=10, kernel="gaussian",
                       show_progress=False, preprocess=False)
-modelss = regr_mts3.fit(df_train, df_test)
+models = regr_mts3.fit(df_train, df_test)
 model_dictionary = regr_mts3.provide_models(df_train, df_test)
 print(models)
 print(models["WINKLERSCORE"])
@@ -81,7 +81,7 @@
                       replications=100, kernel="gaussian",
                       type_pi="scp2-kde",
                       show_progress=False, preprocess=False)
-modelss = regr_mts3.fit(df2_train, df2_test)
+models = regr_mts3.fit(df2_train, df2_test)
 model_dictionary = regr_mts3.provide_models(df2_train, df2_test)
 print(models)
 print(models["WINKLERSCORE"])

examples/lazy_deepmts_horizon.py

@@ -44,7 +44,7 @@
                       show_progress=False, preprocess=False, 
                       estimators=["ElasticNetCV", "RidgeCV"],
                       h=5, )
-models, predictions = regr_mts3.fit(df_train, df_test)
+models = regr_mts3.fit(df_train, df_test)
 model_dictionary = regr_mts3.provide_models(df_train, df_test)
 print(models)
 print(model_dictionary["DeepMTS(ElasticNetCV)"])
@@ -56,7 +56,7 @@
                       n_layers=3,
                       show_progress=False, preprocess=False,
                       h=5, )
-models, predictions = regr_mts.fit(df_train, df_test)
+models = regr_mts.fit(df_train, df_test)
 model_dictionary = regr_mts.provide_models(df_train, df_test)
 print(models)
 print(model_dictionary["DeepMTS(LinearSVR)"])
@@ -69,7 +69,7 @@
                       replications=10, kernel="gaussian",
                       show_progress=False, preprocess=False,
                       h=5, )
-models, predictions = regr_mts3.fit(df_train, df_test)
+models = regr_mts3.fit(df_train, df_test)
 model_dictionary = regr_mts3.provide_models(df_train, df_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])
@@ -83,7 +83,7 @@
                       type_pi="scp2-kde",
                       show_progress=False, preprocess=False,
                       h=5, )
-models, predictions = regr_mts3.fit(df2_train, df2_test)
+models = regr_mts3.fit(df2_train, df2_test)
 model_dictionary = regr_mts3.provide_models(df2_train, df2_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])

examples/lazy_mts.py

@@ -35,7 +35,7 @@
                        type_pi = "kde",
                        replications=100, kernel="gaussian",
                        show_progress=True, preprocess=False)
-models, predictions = regr_mts4.fit(df_train, df_test)
+models = regr_mts4.fit(df_train, df_test)
 model_dictionary = regr_mts4.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
@@ -46,7 +46,7 @@
                       kernel="gaussian",
                       replications=100, 
                       show_progress=True, preprocess=False)
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
@@ -56,7 +56,7 @@
                       kernel="tophat",
                       replications=100, 
                       show_progress=True, preprocess=False)
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
@@ -66,7 +66,7 @@
                       kernel="tophat",
                       replications=100, 
                       show_progress=True, preprocess=False)
-models, predictions = regr_mts6.fit(df_train, df_test)
+models = regr_mts6.fit(df_train, df_test)
 model_dictionary = regr_mts6.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
@@ -90,7 +90,7 @@
                        replications = 100, kernel="gaussian",
                        estimators = ["Ridge", "Lasso", "LarsCV", "LassoCV", "LassoLarsCV"],
                        show_progress=True, preprocess=False)
-models, predictions = regr_mts4.fit(df_train, df_test)
+models = regr_mts4.fit(df_train, df_test)
 model_dictionary = regr_mts4.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
@@ -100,15 +100,15 @@
                         replications=100, kernel="gaussian",
                         estimators = ["Ridge", "Lasso", "LarsCV", "LassoCV", "LassoLarsCV"],
                         show_progress=False, preprocess=False)
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
 regr_mts5 = ns.LazyDeepMTS(verbose=1, ignore_warnings=True, custom_metric=None,
                     lags = 20, n_hidden_features=7, n_clusters=2,
                     type_pi = "gaussian",
                     show_progress=False, preprocess=False)
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 

examples/lazy_mts_horizon.py

@@ -36,7 +36,7 @@
                        replications=100, kernel="gaussian",
                        show_progress=True, preprocess=False,
                        h=5, )
-models, predictions = regr_mts4.fit(df_train, df_test)
+models = regr_mts4.fit(df_train, df_test)
 model_dictionary = regr_mts4.provide_models(df_train, df_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])
@@ -50,7 +50,7 @@
                       replications=100, 
                       show_progress=True, preprocess=False,
                       h=5, )
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])
@@ -64,7 +64,7 @@
                       replications=100, 
                       show_progress=True, preprocess=False,
                       h=5, )
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])
@@ -76,7 +76,7 @@
                       type_pi="scp2-block-bootstrap", 
                       replications=100, show_progress=True, 
                       preprocess=False, h=5 )
-models, predictions = regr_mts6.fit(df_train, df_test)
+models = regr_mts6.fit(df_train, df_test)
 model_dictionary = regr_mts6.provide_models(df_train, df_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])
@@ -102,7 +102,7 @@
                        estimators = ["Ridge", "Lasso", "LarsCV", "LassoCV", "LassoLarsCV"],
                        show_progress=True, preprocess=False,
                          h=5, )
-models, predictions = regr_mts4.fit(df_train, df_test)
+models = regr_mts4.fit(df_train, df_test)
 model_dictionary = regr_mts4.provide_models(df_train, df_test)
 print(models)
 print(models[['WINKLERSCORE', 'COVERAGE']])
@@ -116,7 +116,7 @@
                         estimators = ["Ridge", "Lasso", "LarsCV", "LassoCV", "LassoLarsCV"],
                         show_progress=False, preprocess=False,
                          h=5, )
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models[['WINKLERSCORE', 'COVERAGE']])
 
@@ -127,7 +127,7 @@
                     #type_pi = "gaussian",
                     show_progress=False, preprocess=False,
                     h=5, )
-models, predictions = regr_mts5.fit(df_train, df_test)
+models = regr_mts5.fit(df_train, df_test)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models)
 
@@ -139,7 +139,7 @@
                         estimators = ["Ridge", "Lasso", "LarsCV", "LassoCV", "LassoLarsCV"],
                         show_progress=False, preprocess=False,
                          h=5, )
-models, predictions = regr_mts5.fit(df_train, df_test, per_series=True)
+models = regr_mts5.fit(df_train, df_test, per_series=True)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models)
 
@@ -151,6 +151,6 @@
                     #type_pi = "gaussian",
                     show_progress=False, preprocess=False,
                     h=5, )
-models, predictions = regr_mts5.fit(df_train, df_test, per_series=True)
+models = regr_mts5.fit(df_train, df_test, per_series=True)
 model_dictionary = regr_mts5.provide_models(df_train, df_test)
 print(models)