View the original Github Repo:

Naive Bayes is a popular rstatistical technique for e-mail filtering. It uses bag-of-words features (use countvectorizer to create the features from the text) to identify the email spam.
It uses a Bayes’ theorem to calculate a probability that an email is spam or not and it usually has low false positive spam detection rates, meaning that it rarely misclassify an legitimate email as spam. However, as its name suggest, it assumes strong and naive assumption about the independence of the words, which is rarely the case for words in the context. Click here

Import the library

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

Explore the datasets

The datasets has been collected for SMS Spam research and it has two columns. One columns has the text and another shows the label(ham or spam)

spam_df = pd.read_csv("emails.csv")

spam_df.head()

	text	spam
0	Subject: naturally irresistible your corporate...	1
1	Subject: the stock trading gunslinger fanny i...	1
2	Subject: unbelievable new homes made easy im ...	1
3	Subject: 4 color printing special request add...	1
4	Subject: do not have money , get software cds ...	1

spam_df.tail()

	text	spam
5723	Subject: re : research and development charges...	0
5724	Subject: re : receipts from visit jim , than...	0
5725	Subject: re : enron case study update wow ! a...	0
5726	Subject: re : interest david , please , call...	0
5727	Subject: news : aurora 5 . 2 update aurora ve...	0

spam_df.describe()

	spam
count	5728.000000
mean	0.238827
std	0.426404
min	0.000000
25%	0.000000
50%	0.000000
75%	0.000000
max	1.000000

spam_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5728 entries, 0 to 5727
Data columns (total 2 columns):
 #   Column  Non-Null Count  Dtype 
---  ------  --------------  ----- 
 0   text    5728 non-null   object
 1   spam    5728 non-null   int64 
dtypes: int64(1), object(1)
memory usage: 89.6+ KB

spam_df["spam"].value_counts(normalize=True)

0    0.761173
1    0.238827
Name: spam, dtype: float64

Visualize the datasets

spam_df["length"] = spam_df["text"].apply(len)
spam_df.head()

	text	spam	length
0	Subject: naturally irresistible your corporate...	1	1484
1	Subject: the stock trading gunslinger fanny i...	1	598
2	Subject: unbelievable new homes made easy im ...	1	448
3	Subject: 4 color printing special request add...	1	500
4	Subject: do not have money , get software cds ...	1	235

spam_df["length"].plot(bins=100, kind="hist")

<AxesSubplot:ylabel='Frequency'>

png

spam_df.length.describe()

count     5728.000000
mean      1556.768680
std       2042.649812
min         13.000000
25%        508.750000
50%        979.000000
75%       1894.250000
max      43952.000000
Name: length, dtype: float64

Clean the datasets

#remove punctutation
import string

#remove stopwords
from nltk.corpus import stopwords

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(spam_df["text"], spam_df["spam"], test_size=0.2)

def clean_message(message):
    message = [char for char in message if char not in string.punctuation]
    message = "".join(message)
    message = [word for word in message.split() if word.lower() not in stopwords.words("english")]
    message = " ".join(message) 
    return message

from sklearn.feature_extraction.text import CountVectorizer
CV = CountVectorizer()#min_df=0.01, max_features=300, stop_words="english")

CV.fit(X_train.apply(clean_message))

CountVectorizer()

X_train = CV.transform(X_train)
X_test = CV.transform(X_test)

Training the models

from sklearn.naive_bayes import MultinomialNB

NB_classifier = MultinomialNB()
NB_classifier.fit(X_train, y_train)

MultinomialNB()

Evaluate the model

from sklearn.metrics import classification_report, confusion_matrix

#predict the training set
y_predict_train = NB_classifier.predict(X_train)
cm = confusion_matrix(y_train, y_predict_train)
sns.heatmap(cm, annot=True)

<AxesSubplot:>

png

#predict the test set
y_predict_test = NB_classifier.predict(X_test)
cm = confusion_matrix(y_test, y_predict_test)
sns.heatmap(cm, annot=True)

<AxesSubplot:>

png

print(classification_report(y_test, y_predict_test))

              precision    recall  f1-score   support

           0       1.00      1.00      1.00       878
           1       0.99      0.99      0.99       268

    accuracy                           0.99      1146
   macro avg       0.99      0.99      0.99      1146
weighted avg       0.99      0.99      0.99      1146

← Previous Post Next Post →

Use Naive Bayes to classify spam messages