def str_img(txt, datadir_path):
img = Image.new('RGB', (900, 900), color='white')
if type(txt) == 'dict':
txt = json.dumps(txt)
d = ImageDraw.Draw(img)
d.text((10, 10), str(txt), fill=(0, 0, 0))
output_path = os.path.join(datadir_path, 'pil_text.png')
img.save(output_path)
PImage(output_path)
else:
img = Image.new('RGB', (100, 30), color='white')
d = ImageDraw.Draw(img)
d.text((10, 10), str(txt), fill=(0, 0, 0))
output_path = os.path.join(datadir_path, 'pil_text.png')
img.save(output_path)
PImage(output_path)
def visualize(classifier, X_test):
with open("tree1.dot", 'w') as f:
f = tree.export_graphviz(classifier,
out_file=f,
impurity=False,
feature_names=X_test.columns.values,
class_names=['No', 'Yes'],
rounded=True,
filled=True)
check_call(['dot', '-Tpng', 'tree1.dot', '-o', 'tree1.png'])
img = Image.open('tree1.png')
draw = ImageDraw.Draw(img)
img.save('sample-out.png')
PImage('sample-out.png')
f = tree.export_graphviz(classifier,
out_file=f,
impurity=False,
feature_names=X_test.columns.values,
class_names=['No', 'Yes'],
rounded=True,
filled=True)
# Convert .dot to .png to allow display in web notebook
check_call(['dot', '-Tpng', 'tree1.dot', '-o', 'tree1.png'])
# Annotating chart with PIL
img = Image.open("tree1.png")
draw = ImageDraw.Draw(img)
img.save('sample-out.png')
PImage("sample-out.png")
classifier = DecisionTreeClassifier(max_depth=3)
classifier.fit(X_train, y_train)
print('train score...', accuracy_score(y_train, classifier.predict(X_train)))
print('test score...', accuracy_score(y_test, classifier.predict(X_test)))
with open("tree1.dot", 'w') as f:
f = tree.export_graphviz(classifier,
out_file=f,
impurity=False,
feature_names=X_test.columns.values,
class_names=['No', 'Yes'],
rounded=True,
filled=True)
# Convert .dot to .png to allow display in web notebook
check_call(['dot', '-Tpng', 'tree1.dot', '-o', 'tree1.png'])
# exportar el modelo a archivo .dot
with open(r"tree1.dot", 'w') as f:
f = tree.export_graphviz(decision_tree,
out_file=f,
max_depth=7,
impurity=True,
feature_names=list(
artists_encoded.drop(['top'], axis=1)),
class_names=['No', 'N1 Billboard'],
rounded=True,
filled=True)
# Convertir el archivo .dot a png para poder visualizarlo
check_call(['dot', '-Tpng', r'tree1.dot', '-o', r'tree1.png'])
PImage("tree1.png")
# In[ ]:
acc_decision_tree = round(decision_tree.score(x_train, y_train) * 100, 2)
# In[12]:
#predecir artista CAMILA CABELLO featuring YOUNG THUG
# con su canción Havana llego a numero 1 Billboard US en 2017
#response.json()
#RECIBIR JSON CON ARRAY
try:
data = json.loads(sys.argv[1])
except:
with open("safe-loans.dot", 'w') as f:
f = tree.export_graphviz(clf,
out_file=f,
max_depth=sDepth,
impurity=True,
feature_names=list(X_train),
class_names=['not safe', 'safe'],
rounded=True,
filled=True)
check_call(['dot', '-Tpng', 'safe-loans.dot', '-o', 'safe-loans.png'])
img = Image.open("safe-loans.png")
draw = ImageDraw.Draw(img)
img.save('output.png')
PImage("output.png")
trainPredictions = clf.predict(X_train)
testPredictions = clf.predict(X_test)
fprTrain, tprTrain, thresholdsTrain = roc_curve(Y_train, trainPredictions)
roc_auc_train = auc(fprTrain, tprTrain)
print(roc_auc_train)
fprTest, tprTest, thresholdsTest = roc_curve(Y_test, testPredictions)
roc_auc_test = auc(fprTest, tprTest)
print(roc_auc_test)
pl.clf()
pl.plot(fprTest, tprTrain, label='ROC curve (area = %0.2f)' % roc_auc_test)
pl.plot([0, 1], [0, 1], '-k')
#https://www.kaggle.com/dmilla/introduction-to-decision-trees-titanic-dataset
from IPython.display import Image as PImage
from subprocess import check_call
def drawTree(datree):
with open("tree1.dot", 'w') as f:
f = tree.export_graphviz(datree,
out_file=f,
rounded = True,
filled= True )
#Convert .dot to .png to allow display in web notebook
check_call(['dot','-Tpng','tree1.dot','-o','tree1.png'])
return("tree1.png")
#use the function to draw the tree
PImage(drawTree(dtc))
###some code qualifying the data
# <h3>Let's optimize the pruning level of the code</h3>
# In[53]:
### some optimization code using cross validation!?
# Tree complexity parameter - constraint on maximum depth
tc = np.arange(2, 41, 1)
# K-fold crossvalidation
K = 10
out_file=f,
max_depth=4,
impurity=False,
feature_names=attrition_final.columns.values,
class_names=['No', 'Yes'],
rounded=True,
filled=True)
#Convert .dot to .png to allow display in web notebook
check_call(['dot', '-Tpng', 'tree1.dot', '-o', 'tree1.png'])
# Annotating chart with PIL
img = Image.open("tree1.png")
draw = ImageDraw.Draw(img)
img.save('sample-out.png')
PImage("sample-out.png", height=2000, width=1900)
# Gradient Boosting Parameters
gb_params = {
'n_estimators': 1500,
'max_features': 0.9,
'learning_rate': 0.25,
'max_depth': 4,
'min_samples_leaf': 2,
'subsample': 1,
'max_features': 'sqrt',
'random_state': seed,
'verbose': 0
}
gb = GradientBoostingClassifier(**gb_params)
# Fit the model to our SMOTEd train and target
gb.fit(smote_train, smote_target)
rounded = True,
filled= True )#)
#Convert .dot to .png to allow display in web notebook
check_call(['dot','-Tpng','tree1.dot','-o','tree1.png'])
# Annotating chart with PIL
img = Image.open("tree1.png")
draw = ImageDraw.Draw(img)
font = ImageFont.truetype('/usr/share/fonts/truetype/liberation/LiberationSerif-Bold.ttf', 26)
draw.text((10, 0), # Drawing offset (position)
'"Title <= Income', # Text to draw
(0,0,255), # RGB desired color
font=font) # ImageFont object with desired font
img.save('tree_income.png')
PImage("tree_income.png")
#######
###METRICS DECISION TREE
y_pred=decision_tree.predict(val_X)
print("Accuracy:", decision_tree.score(val_X, val_y))
print("Precision:", precision_score(val_y, y_pred))
print("Recall:", recall_score(val_y, y_pred))
print("F1:", f1_score(val_y, y_pred))
print("Area under precision Recall:", average_precision_score(val_y, y_pred))
##############3
#####3
#######3 RANDOM FOREST##################################################################################################
rounded = True,
filled= True )#)
#Convert .dot to .png to allow display in web notebook
check_call(['dot','-Tpng','treeisfraud.dot','-o','treeisfraud.png'])
# Annotating chart with PIL
img = Image.open("treeisfraud.png")
draw = ImageDraw.Draw(img)
font = ImageFont.truetype('/usr/share/fonts/truetype/liberation/LiberationSerif-Bold.ttf', 26)
draw.text((10, 0), # Drawing offset (position)
'"Title <= Is Fraud', # Text to draw
(0,0,255), # RGB desired color
font=font) # ImageFont object with desired font
img.save('tree_isfraud.png')
PImage("tree_isfraud.png")
########33
####### KNeighborsClassifier ###################################################################################################3
##############
from sklearn.neighbors import KNeighborsClassifier
neiclassifier = KNeighborsClassifier(n_neighbors=5)
neiclassifier.fit(train_X, train_y)
y_pred_nei = neiclassifier.predict(val_X)
KNeighbors=[neiclassifier.score(val_X, val_y),precision_score(val_y, y_pred_nei),recall_score(val_y, y_pred_nei),f1_score(val_y,y_pred_nei)]
results['KNeighbors']=KNeighbors
