import necessary library¶

In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

load data¶

In [2]:
df = pd.read_csv('spam.tsv', sep='\t')
In [3]:
df.head()
Out[3]:
label message length punct
0 ham Go until jurong point, crazy.. Available only ... 111 9
1 ham Ok lar... Joking wif u oni... 29 6
2 spam Free entry in 2 a wkly comp to win FA Cup fina... 155 6
3 ham U dun say so early hor... U c already then say... 49 6
4 ham Nah I don't think he goes to usf, he lives aro... 61 2
In [4]:
df['message'].iloc[0]
Out[4]:
'Go until jurong point, crazy.. Available only in bugis n great world la e buffet... Cine there got amore wat...'
In [5]:
df.tail()
Out[5]:
label message length punct
5567 spam This is the 2nd time we have tried 2 contact u... 160 8
5568 ham Will ü b going to esplanade fr home? 36 1
5569 ham Pity, * was in mood for that. So...any other s... 57 7
5570 ham The guy did some bitching but I acted like i'd... 125 1
5571 ham Rofl. Its true to its name 26 1

exploratory data analysis¶

In [6]:
plt.figure(figsize=(16, 4))
sns.countplot(data=df, x='label')
plt.show()
In [7]:
print(df['label'].value_counts()/len(df))

ham     0.865937
spam    0.134063
Name: label, dtype: float64
In [8]:
df.isna().sum()
Out[8]:
label      0
message    0
length     0
punct      0
dtype: int64

data preprocessing¶

In [9]:
ham = df[df['label'] == 'ham']
spam = df[df['label'] == 'spam']
In [10]:
ham = ham.sample(len(spam))
len(ham), len(spam)
Out[10]:
(747, 747)
In [11]:
df = ham.append(spam, ignore_index=True)

C:\Users\yandiher\AppData\Local\Temp\ipykernel_10108\2145385498.py:1: FutureWarning: The frame.append method is deprecated and will be removed from pandas in a future version. Use pandas.concat instead.
  df = ham.append(spam, ignore_index=True)
In [12]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1494 entries, 0 to 1493
Data columns (total 4 columns):
 #   Column   Non-Null Count  Dtype 
---  ------   --------------  ----- 
 0   label    1494 non-null   object
 1   message  1494 non-null   object
 2   length   1494 non-null   int64 
 3   punct    1494 non-null   int64 
dtypes: int64(2), object(2)
memory usage: 46.8+ KB

building algorithm¶

In [13]:
from sklearn.model_selection import train_test_split, GridSearchCV
In [14]:
x = df['message'].values
y = df['label'].values
xTrain, xTest, yTrain, yTest = train_test_split(x, y, train_size=0.8, random_state=42, stratify=y)
In [15]:
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.pipeline import Pipeline
In [16]:
models = [RandomForestClassifier(),
          LogisticRegression(),
          KNeighborsClassifier(),
          SVC()]

names = ['RandomForestClassifier',
         'LogisticRegression',
         'KNeighborsClassifier',
         'SVC']
In [17]:
scores = []
for model in models:
    clf = Pipeline([('tfidf', TfidfVectorizer()), ('classifier', model)])
    clf.fit(xTrain, yTrain)
    score = clf.score(xTest, yTest)
    scores.append(score)

comparison = pd.DataFrame(data={'names': names, 'scores': scores},
                          columns=['names', 'scores']).sort_values(by=['scores'], ascending=False)
In [18]:
comparison
Out[18]:
names scores
3 SVC 0.966555
0 RandomForestClassifier 0.956522
1 LogisticRegression 0.953177
2 KNeighborsClassifier 0.929766
In [19]:
clf = Pipeline([('tfidf', TfidfVectorizer()), ('classifier', SVC())])
clf.fit(xTrain, yTrain)
prediction = clf.predict(xTest)
In [20]:
from sklearn.metrics import classification_report, ConfusionMatrixDisplay
In [21]:
print(classification_report(yTest, prediction))

              precision    recall  f1-score   support

         ham       0.95      0.99      0.97       150
        spam       0.99      0.95      0.97       149

    accuracy                           0.97       299
   macro avg       0.97      0.97      0.97       299
weighted avg       0.97      0.97      0.97       299
In [22]:
ConfusionMatrixDisplay.from_predictions(yTest, prediction)
plt.title(label='Support Vector Classifier')
plt.show()

grid search¶

In [23]:
vectorizer = TfidfVectorizer()
In [24]:
vectorizer.fit(x)
Out[24]:
TfidfVectorizer()
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.
TfidfVectorizer()
In [25]:
x = vectorizer.transform(x)
y = pd.get_dummies(data=y, drop_first=True)
y = y.values.ravel()
In [26]:
params = {'C': [0.001, 0.01, 0.1, 1],
          'kernel': ['rbf', 'poly', 'linear'],
          'gamma': [0.1, 1, 10, 100]}
In [27]:
grid = GridSearchCV(estimator=SVC(), param_grid=params, cv=5, scoring='accuracy')
grid.fit(x, y)
Out[27]:
GridSearchCV(cv=5, estimator=SVC(),
             param_grid={'C': [0.001, 0.01, 0.1, 1], 'gamma': [0.1, 1, 10, 100],
                         'kernel': ['rbf', 'poly', 'linear']},
             scoring='accuracy')
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.
GridSearchCV(cv=5, estimator=SVC(),
             param_grid={'C': [0.001, 0.01, 0.1, 1], 'gamma': [0.1, 1, 10, 100],
                         'kernel': ['rbf', 'poly', 'linear']},
             scoring='accuracy')
SVC()
SVC()
In [28]:
print(grid.best_params_)
print(grid.best_score_)

{'C': 1, 'gamma': 0.1, 'kernel': 'linear'}
0.9544903593634262
In [29]:
clf = SVC(**grid.best_params_)
clf.fit(x, y)
Out[29]:
SVC(C=1, gamma=0.1, kernel='linear')
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.
SVC(C=1, gamma=0.1, kernel='linear')
In [30]:
test = ['free tickets if you click this button. get your seat now!!!']
test
Out[30]:
['free tickets if you click this button. get your seat now!!!']
In [31]:
test = vectorizer.transform(test)
if clf.predict(test) == 1:
    print('this message is spam')
else:
    print('this message is ham')

this message is spam