10 - Generative Models for Classification Code Demo

Author

Dr. Cheng-Han Yu

1 R implementation

Code

library(ISLR2)
library(MASS)


Attaching package: 'MASS'

The following object is masked from 'package:ISLR2':

    Boston

Code

library(e1071)
data("Default")

Code

lda_fit <- MASS::lda(default ~ balance, data = Default)
lda_pred <- predict(lda_fit, data = Default)
table(lda_pred$class, Default$default, dnn = c("Predicted", "Actual"))

         Actual
Predicted   No  Yes
      No  9643  257
      Yes   24   76

Code

qda_fit <- MASS::qda(default ~ balance, data = Default)
qda_pred <- predict(qda_fit, data = Default)
table(qda_pred$class, Default$default, dnn = c("Predicted", "Actual"))

         Actual
Predicted   No  Yes
      No  9639  246
      Yes   28   87

Code

nb_fit <- e1071::naiveBayes(default ~ balance, data = Default)
nb_pred <- predict(nb_fit, Default)
(nb_conf <- table(nb_pred, Default$default))

       
nb_pred   No  Yes
    No  9639  246
    Yes   28   87

2 Python implementation

Code

import pandas as pd
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis
from sklearn.naive_bayes import GaussianNB
from sklearn.metrics import confusion_matrix

Code

Default = pd.read_csv("../data/Default.csv")
X = Default[['balance']]
y = Default['default']

Code

lda = LinearDiscriminantAnalysis()
lda.fit(X, y)

LinearDiscriminantAnalysis()

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Code

lda_pred = lda.predict(X)
lda_conf_df = pd.DataFrame(
    confusion_matrix(y_true=y, y_pred=lda_pred), 
    index=[f"Actual: {cls}" for cls in lda.classes_],
    columns=[f"Predicted: {cls}" for cls in lda.classes_]
)
lda_conf_df

             Predicted: No  Predicted: Yes
Actual: No            9643              24
Actual: Yes            257              76

Code

qda = QuadraticDiscriminantAnalysis()
qda.fit(X, y)

QuadraticDiscriminantAnalysis()

Code

qda_pred = qda.predict(X)
qda_conf_df = pd.DataFrame(
    confusion_matrix(y, qda_pred), 
    index=[f"Actual: {cls}" for cls in qda.classes_],
    columns=[f"Predicted: {cls}" for cls in qda.classes_]
)
qda_conf_df

             Predicted: No  Predicted: Yes
Actual: No            9639              28
Actual: Yes            246              87

Code

nb = GaussianNB()
nb.fit(X, y)

GaussianNB()

Code

nb_pred = nb.predict(X)
nb_conf_df = pd.DataFrame(
    confusion_matrix(y, nb_pred),
    index=[f"Actual: {cls}" for cls in nb.classes_],
    columns=[f"Predicted: {cls}" for cls in nb.classes_]
)
nb_conf_df

             Predicted: No  Predicted: Yes
Actual: No            9639              28
Actual: Yes            246              87