From b279acb5bfc1623fa33a315a9bf027881e7dd765 Mon Sep 17 00:00:00 2001
From: Baptiste Bauvin <baptiste.bauvin.1@ulaval.ca>
Date: Sat, 21 Oct 2017 19:05:35 -0400
Subject: [PATCH] Added graphs to vizualize feature importances
---
.../MonoviewClassifiers/Adaboost.py | 20 +++--------
.../MonoviewClassifiers/DecisionTree.py | 18 ++--------
.../MonoviewClassifiers/RandomForest.py | 18 ++--------
.../MonoviewClassifiers/SCM.py | 10 +++---
.../MonoviewClassifiers/SGD.py | 2 ++
.../MonoviewClassifiers/SVMLinear.py | 1 +
.../utils/Interpret.py | 35 +++++++++++++++++++
7 files changed, 54 insertions(+), 50 deletions(-)
create mode 100644 Code/MonoMultiViewClassifiers/utils/Interpret.py
diff --git a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/Adaboost.py b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/Adaboost.py
index 66a39edd..b5f75f49 100644
--- a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/Adaboost.py
+++ b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/Adaboost.py
@@ -4,10 +4,13 @@ from sklearn.model_selection import RandomizedSearchCV
from sklearn.tree import DecisionTreeClassifier
from scipy.stats import randint
import numpy as np
-import cPickle
+# import cPickle
+# import matplotlib.pyplot as plt
+# from matplotlib.ticker import FuncFormatter
from .. import Metrics
from ..utils.HyperParameterSearch import genHeatMaps
+from ..utils.Interpret import getFeatureImportance
# Author-Info
__author__ = "Baptiste Bauvin"
@@ -82,18 +85,5 @@ def getConfig(config):
config["1"])
def getInterpret(classifier, directory):
- featureImportances = classifier.feature_importances_
- sortedArgs = np.argsort(-featureImportances)
- featureImportancesSorted = featureImportances[sortedArgs][:50]
- featureIndicesSorted = sortedArgs[:50]
- featuresImportancesDict = dict((featureIndex, featureImportance)
- for featureIndex, featureImportance in enumerate(featureImportances)
- if featureImportance != 0)
- with open(directory+'-feature_importances.pickle', 'wb') as handle:
- cPickle.dump(featuresImportancesDict, handle)
- interpretString = "Feature importances : \n"
- for featureIndex, featureImportance in zip(featureIndicesSorted, featureImportancesSorted):
- if featureImportance>0:
- interpretString+="- Feature index : "+str(featureIndex)+\
- ", feature importance : "+str(featureImportance)+"\n"
+ interpretString = getFeatureImportance(classifier, directory)
return interpretString
\ No newline at end of file
diff --git a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/DecisionTree.py b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/DecisionTree.py
index 152e9e7d..1dbd83ad 100644
--- a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/DecisionTree.py
+++ b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/DecisionTree.py
@@ -4,10 +4,11 @@ from sklearn.model_selection import RandomizedSearchCV
from scipy.stats import randint
import numpy as np
import graphviz
-import cPickle
+# import cPickle
from .. import Metrics
from ..utils.HyperParameterSearch import genHeatMaps
+from ..utils.Interpret import getFeatureImportance
# Author-Info
__author__ = "Baptiste Bauvin"
@@ -92,18 +93,5 @@ def getInterpret(classifier, directory):
dot_data = tree.export_graphviz(classifier, out_file=None)
graph = graphviz.Source(dot_data)
graph.render(directory+"-tree.pdf")
- featureImportances = classifier.feature_importances_
- sortedArgs = np.argsort(-featureImportances)
- featureImportancesSorted = featureImportances[sortedArgs][:50]
- featureIndicesSorted = sortedArgs[:50]
- featuresImportancesDict = dict((featureIndex, featureImportance)
- for featureIndex, featureImportance in enumerate(featureImportances)
- if featureImportance != 0)
- with open(directory + '-feature_importances.pickle', 'wb') as handle:
- cPickle.dump(featuresImportancesDict, handle)
- interpretString = "Feature importances : \n"
- for featureIndex, featureImportance in zip(featureIndicesSorted, featureImportancesSorted):
- if featureImportance > 0:
- interpretString += "- Feature index : " + str(featureIndex) + \
- ", feature importance : " + str(featureImportance) + "\n"
+ interpretString = getFeatureImportance(classifier, directory)
return interpretString
\ No newline at end of file
diff --git a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/RandomForest.py b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/RandomForest.py
index 66caa2d4..79ea71e4 100644
--- a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/RandomForest.py
+++ b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/RandomForest.py
@@ -3,10 +3,11 @@ from sklearn.pipeline import Pipeline
from sklearn.model_selection import RandomizedSearchCV
from scipy.stats import randint
import numpy as np
-import cPickle
+# import cPickle
from .. import Metrics
from ..utils.HyperParameterSearch import genHeatMaps
+from ..utils.Interpret import getFeatureImportance
# Author-Info
__author__ = "Baptiste Bauvin"
@@ -90,18 +91,5 @@ def getConfig(config):
def getInterpret(classifier, directory):
- featureImportances = classifier.feature_importances_
- sortedArgs = np.argsort(-featureImportances)
- featureImportancesSorted = featureImportances[sortedArgs][:50]
- featureIndicesSorted = sortedArgs[:50]
- featuresImportancesDict = dict((featureIndex, featureImportance)
- for featureIndex, featureImportance in enumerate(featureImportances)
- if featureImportance != 0)
- with open(directory+'-feature_importances.pickle', 'wb') as handle:
- cPickle.dump(featuresImportancesDict, handle)
- interpretString = "Feature importances : \n"
- for featureIndex, featureImportance in zip(featureIndicesSorted, featureImportancesSorted):
- if featureImportance>0:
- interpretString+="- Feature index : "+str(featureIndex)+ \
- ", feature importance : "+str(featureImportance)+"\n"
+ interpretString = getFeatureImportance(classifier, directory)
return interpretString
diff --git a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SCM.py b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SCM.py
index ce998ed4..56084400 100644
--- a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SCM.py
+++ b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SCM.py
@@ -78,10 +78,6 @@ def paramsToSet(nIter, randomState):
return paramsSet
-def getInterpret(classifier, directory):
- return ""
-
-
def getKWARGS(kwargsList):
kwargsDict = {}
for (kwargName, kwargValue) in kwargsList:
@@ -133,4 +129,8 @@ def getConfig(config):
str(config[2])
except:
return "\n\t\t- SCM with model_type: " + config["0"] + ", max_rules : " + str(config["1"]) + ", p : " + \
- str(config["2"])
\ No newline at end of file
+ str(config["2"])
+
+
+def getInterpret(classifier, directory):
+ return ""
diff --git a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SGD.py b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SGD.py
index e3182787..0bbd424d 100644
--- a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SGD.py
+++ b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SGD.py
@@ -91,4 +91,6 @@ def getConfig(config):
"1"] + ", alpha : " + str(config["2"])
def getInterpret(classifier, directory):
+ # TODO : coeffs
return ""
+#
\ No newline at end of file
diff --git a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SVMLinear.py b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SVMLinear.py
index 63872d5f..87ad608a 100644
--- a/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SVMLinear.py
+++ b/Code/MonoMultiViewClassifiers/MonoviewClassifiers/SVMLinear.py
@@ -75,4 +75,5 @@ def getConfig(config):
return "\n\t\t- SVM Linear with C : " + str(config["0"])
def getInterpret(classifier, directory):
+ # TODO : coeffs
return ""
diff --git a/Code/MonoMultiViewClassifiers/utils/Interpret.py b/Code/MonoMultiViewClassifiers/utils/Interpret.py
new file mode 100644
index 00000000..6f1882a5
--- /dev/null
+++ b/Code/MonoMultiViewClassifiers/utils/Interpret.py
@@ -0,0 +1,35 @@
+import numpy as np
+import matplotlib.pyplot as plt
+from matplotlib.ticker import FuncFormatter
+import cPickle
+
+
+def percent(x, pos):
+ 'The two args are the value and tick position'
+ return '%1.1f %%' % (x * 100)
+
+
+def getFeatureImportance(classifier, directory, interpretString=""):
+ featureImportances = classifier.feature_importances_
+ sortedArgs = np.argsort(-featureImportances)
+ featureImportancesSorted = featureImportances[sortedArgs][:50]
+ featureIndicesSorted = sortedArgs[:50]
+ fig, ax = plt.subplots()
+ x = np.arange(50)
+ formatter = FuncFormatter(percent)
+ ax.yaxis.set_major_formatter(formatter)
+ plt.bar(x, featureImportancesSorted)
+ plt.title("Importance depending on feature")
+ fig.savefig(directory + "-feature_importances.png")
+ plt.close()
+ featuresImportancesDict = dict((featureIndex, featureImportance)
+ for featureIndex, featureImportance in enumerate(featureImportances)
+ if featureImportance != 0)
+ with open(directory+'-feature_importances.pickle', 'wb') as handle:
+ cPickle.dump(featuresImportancesDict, handle)
+ interpretString += "Feature importances : \n"
+ for featureIndex, featureImportance in zip(featureIndicesSorted, featureImportancesSorted):
+ if featureImportance>0:
+ interpretString+="- Feature index : "+str(featureIndex)+\
+ ", feature importance : "+str(featureImportance)+"\n"
+ return interpretString
\ No newline at end of file
--
GitLab