From e5fbb2f9c03d025d059858fb78ba73191c9d3c62 Mon Sep 17 00:00:00 2001
From: Baptiste Bauvin <baptiste.bauvin@lis-lab.fr>
Date: Tue, 28 Feb 2023 07:23:15 -0500
Subject: [PATCH] Added some clfs"
---
.../monoview_classifiers/decision_tree.py | 9 +-
.../monoview_classifiers/repeatboost.py | 127 ++++++++++++++++++
.../monoview_classifiers/samba.py | 5 +-
.../monoview_classifiers/scm.py | 30 ++++-
.../monoview_classifiers/scmboost.py | 93 +++++++++++++
.../multiview_classifiers/early_fusion_knn.py | 23 ++++
6 files changed, 278 insertions(+), 9 deletions(-)
create mode 100644 summit/multiview_platform/monoview_classifiers/repeatboost.py
create mode 100644 summit/multiview_platform/monoview_classifiers/scmboost.py
create mode 100644 summit/multiview_platform/multiview_classifiers/early_fusion_knn.py
diff --git a/summit/multiview_platform/monoview_classifiers/decision_tree.py b/summit/multiview_platform/monoview_classifiers/decision_tree.py
index f8392df4..00c56dee 100644
--- a/summit/multiview_platform/monoview_classifiers/decision_tree.py
+++ b/summit/multiview_platform/monoview_classifiers/decision_tree.py
@@ -18,21 +18,22 @@ class DecisionTree(DecisionTreeClassifier, BaseMonoviewClassifier):
"""
def __init__(self, random_state=None, max_depth=None,
- criterion='gini', splitter='best', **kwargs):
+ criterion='gini', splitter='best', class_weight='balanced',
+ **kwargs):
DecisionTreeClassifier.__init__(self,
max_depth=max_depth,
criterion=criterion,
- class_weight="balanced",
splitter=splitter,
+ class_weight=class_weight,
random_state=random_state
)
- self.param_names = ["max_depth", "criterion", "splitter",
+ self.param_names = ["max_depth", "criterion", "splitter", "class_weight",
'random_state']
self.classed_params = []
self.distribs = [CustomRandint(low=1, high=300),
["gini", "entropy"],
- ["best", "random"], [random_state]]
+ ["best", "random"],["balanced", None] ,[random_state]]
self.weird_strings = {}
def get_interpretation(self, directory, base_file_name, y_test, feature_ids,
diff --git a/summit/multiview_platform/monoview_classifiers/repeatboost.py b/summit/multiview_platform/monoview_classifiers/repeatboost.py
new file mode 100644
index 00000000..cec9230c
--- /dev/null
+++ b/summit/multiview_platform/monoview_classifiers/repeatboost.py
@@ -0,0 +1,127 @@
+import os
+import time
+
+import numpy as np
+from sklearn.ensemble import AdaBoostClassifier
+
+
+from .. import metrics
+from ..monoview.monoview_utils import BaseMonoviewClassifier, get_accuracy_graph
+from summit.multiview_platform.utils.hyper_parameter_search import CustomRandint
+from ..monoview_classifiers.decision_tree import DecisionTree
+from ..utils.base import base_boosting_estimators
+
+# Author-Info
+__author__ = "Baptiste Bauvin"
+__status__ = "Prototype" # Production, Development, Prototype
+
+classifier_class_name = "REPboost"
+
+
+class RepDT(DecisionTree):
+
+ def fit(self, X, y, sample_weight=None, check_input=True):
+ if sample_weight is not None:
+ new_X, new_y = self.fake_repetitions(X, y, sample_weight,
+ precision=5)
+ else:
+ new_X = X
+ new_y = y
+ DecisionTree.fit(self, new_X, new_y)
+ return self
+
+ def fake_repetitions(self, X, y, sample_weight, precision=3):
+ sample_repetitions = (np.round(sample_weight,
+ precision) * 10 ** precision).astype(
+ np.int64)
+ for ind, sample_rep in enumerate(sample_repetitions):
+ if sample_rep == 0:
+ sample_repetitions[ind] = 1
+ gcd = np.gcd.reduce(sample_repetitions)
+ sample_repetitions = (sample_repetitions / gcd).astype(np.int64)
+ new_X = np.zeros(
+ (X.shape[0]+ int(np.sum(sample_repetitions)-len(sample_repetitions)), X.shape[1]))
+ new_y = np.zeros(X.shape[0]+ int(np.sum(sample_repetitions)-len(sample_repetitions)))
+ ind = 0
+ for sample_index, (sample_rep, sample, label) in enumerate(
+ zip(sample_repetitions, X, y)):
+ new_X[ind:ind + sample_rep, :] = sample
+ new_y[ind:ind + sample_rep] = label
+ ind += sample_rep
+ return new_X, new_y
+
+class REPboost(AdaBoostClassifier, BaseMonoviewClassifier):
+ """
+ This class is an adaptation of scikit-learn's `AdaBoostClassifier <https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.AdaBoostClassifier.html#sklearn.ensemble.AdaBoostClassifier>`_
+
+ """
+
+ def __init__(self, random_state=None, n_estimators=50,
+ base_estimator=RepDT(max_depth=1),
+ base_estimator_config=None, **kwargs):
+ # base_estimator = BaseMonoviewClassifier.get_base_estimator(self,
+ # base_estimator,
+ # base_estimator_config)
+ AdaBoostClassifier.__init__(self,
+ random_state=random_state,
+ n_estimators=n_estimators,
+ base_estimator=base_estimator,
+ algorithm="SAMME"
+ )
+ self.param_names = ["n_estimators",]
+ self.classed_params = []
+ self.distribs = [CustomRandint(low=1, high=100),]
+ self.weird_strings = {}
+ self.plotted_metric = metrics.zero_one_loss
+ self.plotted_metric_name = "zero_one_loss"
+ self.base_estimator_config = base_estimator_config
+ self.step_predictions = None
+
+ def fit(self, X, y, sample_weight=None):
+ begin = time.time()
+ AdaBoostClassifier.fit(self, X, y)
+ end = time.time()
+ self.train_time = end - begin
+ self.train_shape = X.shape
+ self.base_predictions = np.array(
+ [estim.predict(X) for estim in self.estimators_])
+ self.metrics = np.array([self.plotted_metric.score(pred, y) for pred in
+ self.staged_predict(X)])
+ return self
+
+ def predict(self, X):
+ begin = time.time()
+ pred = AdaBoostClassifier.predict(self, X)
+ end = time.time()
+ self.pred_time = end - begin
+ self.step_predictions = np.array(
+ [step_pred for step_pred in self.staged_predict(X)])
+ return pred
+
+ def get_interpretation(self, directory, base_file_name, y_test, feature_ids,
+ multi_class=False): # pragma: no cover
+ interpretString = ""
+ # interpretString += self.get_feature_importance(directory,
+ # base_file_name,
+ # feature_ids)
+ # interpretString += "\n\n Estimator error | Estimator weight\n"
+ # interpretString += "\n".join(
+ # [str(error) + " | " + str(weight / sum(self.estimator_weights_)) for
+ # error, weight in
+ # zip(self.estimator_errors_, self.estimator_weights_)])
+ # step_test_metrics = np.array(
+ # [self.plotted_metric.score(y_test, step_pred) for step_pred in
+ # self.step_predictions])
+ # get_accuracy_graph(step_test_metrics, "Adaboost",
+ # os.path.join(directory,
+ # base_file_name + "test_metrics.png"),
+ # self.plotted_metric_name, set="test")
+ # np.savetxt(os.path.join(directory, base_file_name + "test_metrics.csv"),
+ # step_test_metrics,
+ # delimiter=',')
+ # np.savetxt(
+ # os.path.join(directory, base_file_name + "train_metrics.csv"),
+ # self.metrics, delimiter=',')
+ # np.savetxt(os.path.join(directory, base_file_name + "times.csv"),
+ # np.array([self.train_time, self.pred_time]), delimiter=',')
+ return interpretString
diff --git a/summit/multiview_platform/monoview_classifiers/samba.py b/summit/multiview_platform/monoview_classifiers/samba.py
index a69de937..a2ca6a30 100644
--- a/summit/multiview_platform/monoview_classifiers/samba.py
+++ b/summit/multiview_platform/monoview_classifiers/samba.py
@@ -61,7 +61,7 @@ class SamBAClf(NeighborHoodClassifier, BaseMonoviewClassifier):
"relevance",
"distance",
"train_weighting", "b", "pred_train", "normalizer",
- "normalize_dists", "a", "class_weight"]
+ "normalize_dists", "a", "class_weight", "base_estimator"]
self.distribs = [CustomRandint(low=1, high=70),
[ExpRelevance()],
[EuclidianDist(), PolarDist(), ExpEuclidianDist(), Jaccard()],
@@ -70,7 +70,8 @@ class SamBAClf(NeighborHoodClassifier, BaseMonoviewClassifier):
[True, False],
[RobustScaler()],
[True], CustomRandint(0, 10, 'e-'),
- ["balanced", None]]
+ ["balanced", None],
+ ]
self.classed_params = []
self.weird_strings = {}
diff --git a/summit/multiview_platform/monoview_classifiers/scm.py b/summit/multiview_platform/monoview_classifiers/scm.py
index d094ceec..4ee442f1 100644
--- a/summit/multiview_platform/monoview_classifiers/scm.py
+++ b/summit/multiview_platform/monoview_classifiers/scm.py
@@ -63,9 +63,14 @@ class SCM(scm, BaseMonoviewClassifier):
self.classed_params = []
self.weird_strings = {}
- def fit(self, X, y, tiebreaker=None, iteration_callback=None, **fit_params):
- self.n_features = X.shape[1]
- scm.fit(self, X, y, tiebreaker=None, iteration_callback=None, **fit_params)
+ def fit(self, X, y, tiebreaker=None, iteration_callback=None, sample_weight=None, **fit_params):
+ if sample_weight is not None:
+ new_X, new_y = self.fake_repetitions(X, y, sample_weight, precision=5)
+ else:
+ new_X = X
+ new_y = y
+ self.n_features = new_X.shape[1]
+ scm.fit(self, new_X, new_y, tiebreaker=None, iteration_callback=None, **fit_params)
self.feature_importances_ = np.zeros(self.n_features)
# sum the rules importances :
# rules_importances = estim.get_rules_importances() #activate it when pyscm will implement importance
@@ -76,6 +81,25 @@ class SCM(scm, BaseMonoviewClassifier):
self.feature_importances_ /= np.sum(self.feature_importances_)
return self
+ def fake_repetitions(self, X, y, sample_weight, precision=3):
+ sample_repetitions = (np.round(sample_weight, precision)*10**precision).astype(np.int64)
+ for ind, sample_rep in enumerate(sample_repetitions):
+ if sample_rep==0:
+ sample_repetitions[ind] = 1
+ gcd = np.gcd.reduce(sample_repetitions)
+ sample_repetitions = (sample_repetitions/gcd).astype(np.int64)
+ new_X = np.zeros((X.shape[0]+ int(np.sum(sample_repetitions)-len(sample_repetitions)), X.shape[1]))
+ new_y = np.zeros(X.shape[0]+ int(np.sum(sample_repetitions)-len(sample_repetitions)))
+ ind=0
+ for sample_index, (sample_rep, sample, label) in enumerate(zip(sample_repetitions, X, y)):
+ new_X[ind:ind+sample_rep, :] = sample
+ new_y[ind:ind+sample_rep] =label
+ ind+=sample_rep
+ return new_X, new_y
+
+
+
+
# def canProbas(self):
# """
# Used to know if the classifier can return label probabilities
diff --git a/summit/multiview_platform/monoview_classifiers/scmboost.py b/summit/multiview_platform/monoview_classifiers/scmboost.py
new file mode 100644
index 00000000..73db0419
--- /dev/null
+++ b/summit/multiview_platform/monoview_classifiers/scmboost.py
@@ -0,0 +1,93 @@
+import os
+import time
+
+import numpy as np
+from sklearn.ensemble import AdaBoostClassifier
+
+
+from .. import metrics
+from ..monoview.monoview_utils import BaseMonoviewClassifier, get_accuracy_graph
+from summit.multiview_platform.utils.hyper_parameter_search import CustomRandint, CustomUniform
+from ..monoview_classifiers.scm import SCM
+from ..utils.base import base_boosting_estimators
+
+# Author-Info
+__author__ = "Baptiste Bauvin"
+__status__ = "Prototype" # Production, Development, Prototype
+
+classifier_class_name = "SCMboost"
+
+
+class SCMboost(AdaBoostClassifier, BaseMonoviewClassifier):
+ """
+ This class is an adaptation of scikit-learn's `AdaBoostClassifier <https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.AdaBoostClassifier.html#sklearn.ensemble.AdaBoostClassifier>`_
+
+ """
+
+ def __init__(self, random_state=None, n_estimators=50,
+ base_estimator=SCM(p=0.49, max_rules=1, model_type="disjunction"),
+ base_estimator_config=None, **kwargs):
+ if "base_estimator__p" in kwargs:
+ base_estimator.p = kwargs["base_estimator__p"]
+ AdaBoostClassifier.__init__(self,
+ random_state=random_state,
+ n_estimators=n_estimators,
+ base_estimator=base_estimator,
+ algorithm="SAMME",)
+ self.param_names = ["n_estimators", "base_estimator__p"]
+ self.classed_params = []
+ self.distribs = [CustomRandint(low=1, high=100), CustomUniform(loc=0, state=1)]
+ self.weird_strings = {}
+ self.plotted_metric = metrics.zero_one_loss
+ self.plotted_metric_name = "zero_one_loss"
+ self.base_estimator_config = base_estimator_config
+ self.step_predictions = None
+
+ def fit(self, X, y, sample_weight=None):
+ begin = time.time()
+ AdaBoostClassifier.fit(self, X, y)
+ end = time.time()
+ self.train_time = end - begin
+ self.train_shape = X.shape
+ self.base_predictions = np.array(
+ [estim.predict(X) for estim in self.estimators_])
+ self.metrics = np.array([self.plotted_metric.score(pred, y) for pred in
+ self.staged_predict(X)])
+ return self
+
+ def predict(self, X):
+ begin = time.time()
+ pred = AdaBoostClassifier.predict(self, X)
+ end = time.time()
+ self.pred_time = end - begin
+ self.step_predictions = np.array(
+ [step_pred for step_pred in self.staged_predict(X)])
+ return pred
+
+ def get_interpretation(self, directory, base_file_name, y_test, feature_ids,
+ multi_class=False): # pragma: no cover
+ interpretString = ""
+ # interpretString += self.get_feature_importance(directory,
+ # base_file_name,
+ # feature_ids)
+ # interpretString += "\n\n Estimator error | Estimator weight\n"
+ # interpretString += "\n".join(
+ # [str(error) + " | " + str(weight / sum(self.estimator_weights_)) for
+ # error, weight in
+ # zip(self.estimator_errors_, self.estimator_weights_)])
+ # step_test_metrics = np.array(
+ # [self.plotted_metric.score(y_test, step_pred) for step_pred in
+ # self.step_predictions])
+ # get_accuracy_graph(step_test_metrics, "Adaboost",
+ # os.path.join(directory,
+ # base_file_name + "test_metrics.png"),
+ # self.plotted_metric_name, set="test")
+ # np.savetxt(os.path.join(directory, base_file_name + "test_metrics.csv"),
+ # step_test_metrics,
+ # delimiter=',')
+ # np.savetxt(
+ # os.path.join(directory, base_file_name + "train_metrics.csv"),
+ # self.metrics, delimiter=',')
+ # np.savetxt(os.path.join(directory, base_file_name + "times.csv"),
+ # np.array([self.train_time, self.pred_time]), delimiter=',')
+ return interpretString
diff --git a/summit/multiview_platform/multiview_classifiers/early_fusion_knn.py b/summit/multiview_platform/multiview_classifiers/early_fusion_knn.py
new file mode 100644
index 00000000..63b37853
--- /dev/null
+++ b/summit/multiview_platform/multiview_classifiers/early_fusion_knn.py
@@ -0,0 +1,23 @@
+from .additions.early_fusion_from_monoview import BaseEarlyFusion
+from ..utils.hyper_parameter_search import CustomUniform, CustomRandint
+
+classifier_class_name = "EarlyFusionSVMRBF"
+
+
+class EarlyFusionSVMRBF(BaseEarlyFusion):
+
+ def __init__(self, random_state=None, n_neighbors=5,
+ weights='uniform', algorithm='auto', p=2, **kwargs):
+ BaseEarlyFusion.__init__(self, random_state=random_state,
+ monoview_classifier="knn",n_neighbors=n_neighbors,
+ weights=weights,
+ algorithm=algorithm,
+ p=p, **kwargs)
+ self.param_names = ["n_neighbors", "weights", "algorithm", "p",
+ "random_state", ]
+ self.classed_params = []
+ self.distribs = [CustomRandint(low=1, high=10), ["uniform", "distance"],
+ ["auto", "ball_tree", "kd_tree", "brute"], [1, 2],
+ [random_state]]
+ self.weird_strings = {}
+ self.random_state = random_state
\ No newline at end of file
--
GitLab