Keras ROC different from Scikit ROC? - machine-learning

Question

From the code below, it looks like evaluating the roc with keras and with scikit actually makes a difference. Does anybody know an explanation?
import tensorflow as tf
from keras.layers import Dense, Input, Dropout
from keras import Sequential
import keras
from keras.constraints import maxnorm
from sklearn.metrics import roc_auc_score
# training data: X_train, y_train
# validation data: X_valid, y_valid
# Define the custom callback we will be using to evaluate roc with scikit
class MyCustomCallback(tf.keras.callbacks.Callback):
def on_epoch_end(self,epoch, logs=None):
y_pred = model.predict(X_valid)
print("roc evaluated with scikit = ",roc_auc_score(y_valid, y_pred))
return
# Define the model.
def model():
METRICS = [
tf.keras.metrics.BinaryAccuracy(name='accuracy'),
tf.keras.metrics.AUC(name='auc'),
]
optimizer="adam"
dropout=0.1
init='uniform'
nbr_features= vocab_size-1 #2500
dense_nparams=256
model = Sequential()
model.add(Dense(dense_nparams, activation='relu', input_shape=(nbr_features,), kernel_initializer=init, kernel_constraint=maxnorm(3)))
model.add(Dropout(dropout))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='binary_crossentropy', optimizer=optimizer,metrics = METRICS)
return model
# instantiate the model
model = model()
# fit the model
history = model.fit(x=X_train, y=y_train, batch_size = 8, epochs = 8, verbose=1,validation_data = (X_valid,y_valid), callbacks=[MyCustomCallback()], shuffle=True, validation_freq=1, max_queue_size=10, workers=4, use_multiprocessing=True)
Output:
Train on 4000 samples, validate on 1000 samples
Epoch 1/8
4000/4000 [==============================] - 15s 4ms/step - loss: 0.7950 - accuracy: 0.7149 - auc: 0.7213 - val_loss: 0.7551 - val_accuracy: 0.7608 - val_auc: 0.7770
roc evaluated with scikit = 0.78766515781747
Epoch 2/8
4000/4000 [==============================] - 15s 4ms/step - loss: 0.0771 - accuracy: 0.8235 - auc: 0.8571 - val_loss: 1.0803 - val_accuracy: 0.8574 - val_auc: 0.8954
roc evaluated with scikit = 0.7795984218252997
Epoch 3/8
4000/4000 [==============================] - 14s 4ms/step - loss: 0.0085 - accuracy: 0.8762 - auc: 0.9162 - val_loss: 1.2084 - val_accuracy: 0.8894 - val_auc: 0.9284
roc evaluated with scikit = 0.7705172905961992
Epoch 4/8
4000/4000 [==============================] - 14s 4ms/step - loss: 0.0025 - accuracy: 0.8982 - auc: 0.9361 - val_loss: 1.1700 - val_accuracy: 0.9054 - val_auc: 0.9424
roc evaluated with scikit = 0.7808804338960933
Epoch 5/8
4000/4000 [==============================] - 14s 4ms/step - loss: 0.0020 - accuracy: 0.9107 - auc: 0.9469 - val_loss: 1.1887 - val_accuracy: 0.9150 - val_auc: 0.9501
roc evaluated with scikit = 0.7811174659489438
Epoch 6/8
4000/4000 [==============================] - 14s 4ms/step - loss: 0.0018 - accuracy: 0.9184 - auc: 0.9529 - val_loss: 1.2036 - val_accuracy: 0.9213 - val_auc: 0.9548
roc evaluated with scikit = 0.7822898825544409
Epoch 7/8
4000/4000 [==============================] - 14s 4ms/step - loss: 0.0017 - accuracy: 0.9238 - auc: 0.9566 - val_loss: 1.2231 - val_accuracy: 0.9258 - val_auc: 0.9579
roc evaluated with scikit = 0.7817036742516923
Epoch 8/8
4000/4000 [==============================] - 14s 4ms/step - loss: 0.0016 - accuracy: 0.9278 - auc: 0.9592 - val_loss: 1.2426 - val_accuracy: 0.9293 - val_auc: 0.9600
roc evaluated with scikit = 0.7817419052279585
As you may see, from epoch 2 onwards keras' and scikit's validation ROCs begin diverging. The same happens if I fit the model and then use keras' model.evaluate(X_valid, y_valid). Any help is greatly appreciated.
EDIT: testing the model on a separate test set, I get roc =0.76 so scikit seems to give the correct answer ( btw X_train has 4000 entries, X_valid has 1000 and test has 15000, quite an unconventional splitting but it is forced by external factors).
Also, suggestions on how to improve performance are equally appreciated.
EDIT2: To answer the reply by #arpitrathi, i modified the callbak but unfortunately without success:
class MyCustomCallback(tf.keras.callbacks.Callback):
def on_epoch_end(self,epoch, logs=None):
y_pred = model.predict_proba(X_valid)
print("roc evaluated with scikit = ",roc_auc_score(y_valid, y_pred))
return
model = model()
history = model.fit(x=X_trainl, y=y_train, batch_size = 8, epochs = 3, verbose=1,validation_data = (X_valid,y_valid), callbacks=[MyCustomCallback()], shuffle=True, validation_freq=1, max_queue_size=10, workers=4, use_multiprocessing=True)
Train on 4000 samples, validate on 1000 samples
Epoch 1/3
4000/4000 [==============================] - 20s 5ms/step - loss: 0.8266 - accuracy: 0.7261 - auc: 0.7409 - val_loss: 0.7547 - val_accuracy: 0.7627 - val_auc: 0.7881
roc evaluated with scikit = 0.7921764130168828
Epoch 2/3
4000/4000 [==============================] - 15s 4ms/step - loss: 0.0482 - accuracy: 0.8270 - auc: 0.8657 - val_loss: 1.0831 - val_accuracy: 0.8620 - val_auc: 0.9054
roc evaluated with scikit = 0.78525915504445
Epoch 3/3
4000/4000 [==============================] - 15s 4ms/step - loss: 0.0092 - accuracy: 0.8794 - auc: 0.9224 - val_loss: 1.2226 - val_accuracy: 0.8928 - val_auc: 0.9340
roc evaluated with scikit = 0.7705555215724655
Also, if I plot training and validation accuracy, i see that they both rapidly converge to 1. Is it strange?

Answer 1

The problem lies in the arguments that you passed to the sklearn function for roc_auc_score() calculation. You should use model.predict_proba() instead of model.predict().
def on_epoch_end(self,epoch, logs=None):
y_pred = model.predict_proba(X_valid)
print("roc evaluated with scikit = ",roc_auc_score(y_valid, y_pred))
return

Answer 2

Sklearn and keras use different default parameters when computing AUC. Increasing the number of thresholds keras uses to compute AUC (i.e., increasing num_thresholds) can help the keras AUC better match the sklearn AUC.