epochs=5,
verbose=1,
batch_size=32,
validation_data=(padded_test_docs, y_test))
# In[81]:
class_names = [
'Computer Science', 'Physics', 'Mathematics', 'Statistics',
'Quantitative Biology', 'Quantitative Finance'
]
predsTest = pd.DataFrame(model.predict(padded_test_docs), columns=class_names)
predsTest['pred'] = predsTest.idxmax(axis=1)
predsTest.head(3)
# In[83]:
roundedPredsTest = predsTest.pred
print('Confusion Matrix: ')
cf_matrix = confusion_matrix(val[val.index.isin(test_index)]['LABEL'],
roundedPredsTest,
labels=class_names)
print(cf_matrix)
meplot.cf_matrix_heatmap(cf_matrix, class_names, 13)
# In[84]:
print(
classification_report(val[val.index.isin(test_index)]['LABEL'],
roundedPredsTest))
plt.plot(epochs, loss, color='r', label='Training loss')
plt.plot(epochs, val_loss, color='b', label='Validation loss')
plt.title('Training and Validation loss')
plt.legend()
plt.show()
# Evaluate model in testing dataset
# In[74]:
predsTest = model.predict(padded_test_docs)
roundedPredsTest = np.round(predsTest)
print('Confusion Matrix: Positive is class 1 and Negative is class 0')
cf_matrix = confusion_matrix(y_test, roundedPredsTest, labels=[1, 0])
print(cf_matrix)
meplot.cf_matrix_heatmap(cf_matrix)
# In[75]:
print(classification_report(y_test, roundedPredsTest))
# ### Explainability
# #### Lime
# In[77]:
from lime import lime_text
from lime.lime_text import LimeTextExplainer
