import numpy as np

import pandas as pd

ds=pd.read_csv('spam.csv',encoding="ISO-8859-1", engine="python")

ds.sample(5)

v1 v2 Unnamed:
2 \
1816 ham Are you going to write ccna exam this week??
NaN
1398 ham No did you check? I got his detailed message now
NaN
4874 ham I know dat feelin had it with Pete! Wuld get w...
NaN
3924 ham As if i wasn't having enough trouble sleeping.
NaN
3368 ham Sorry i've not gone to that place. I.ll do so ...
NaN

Unnamed: 3 Unnamed: 4
1816 NaN NaN
1398 NaN NaN
4874 NaN NaN
3924 NaN NaN
3368 NaN NaN

ds.shape

(5572, 5)

#data cleaning
#EDA(exploratry data analysis)
#text preprocessing
#model building
#evalution of model
#improvement

1. Data Cleaning
ds.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5572 entries, 0 to 5571
Data columns (total 5 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
 0 v1 5572 non-null object
1 v2 5572 non-null object
2 Unnamed: 2 50 non-null object
3 Unnamed: 3 12 non-null object
4 Unnamed: 4 6 non-null object
dtypes: object(5)
memory usage: 217.8+ KB

# no use of last 3 column so drop it

ds.drop(columns=['Unnamed: 2','Unnamed: 3','Unnamed: 4'],inplace=True)

ds.head()

v1 v2
0 ham Go until jurong point, crazy.. Available only ...
1 ham Ok lar... Joking wif u oni...
2 spam Free entry in 2 a wkly comp to win FA Cup fina...
3 ham U dun say so early hor... U c already then say...
4 ham Nah I don't think he goes to usf, he lives aro...

# rename column name

ds.rename(columns={'v1':'target','v2':'text'},inplace=True)
ds.sample(5)

target text
2642 ham They can try! They can get lost, in fact. Tee hee
5281 ham And how you will do that, princess? :)
3224 ham I need... Coz i never go before
1631 ham We not watching movie already. Xy wants 2 shop...
2913 spam Sorry! U can not unsubscribe yet. THE MOB offe...

# level encoding (0,1) in target to get better understanding

from sklearn.preprocessing import LabelEncoder
encoder=LabelEncoder()
encod=encoder.fit_transform(ds['target'])

# add this encoding in target column

ds['target']=encod

ds.head()

target text
0 0 Go until jurong point, crazy.. Available only ...
1 0 Ok lar... Joking wif u oni...
2 1 Free entry in 2 a wkly comp to win FA Cup fina...
3 0 U dun say so early hor... U c already then say...
4 0 Nah I don't think he goes to usf, he lives aro...

# check is there any missing value

ds.isnull().sum()
target 0
text 0
dtype: int64

# check for duplicaate values

ds.duplicated().sum()

403

# remove duplicates
ds=ds.drop_duplicates(keep='first')

ds.duplicated().sum()
#no duplicate value

ds.shape

(5169, 2)

EDA ##exploratory data analysis

#count how many spam value or not spam value
ds['target'].value_counts()

0 4516
1 653
Name: target, dtype: int64

# repersent in piechart
import matplotlib.pyplot as plt
plt.pie(ds['target'].value_counts(), labels=['non
spam','spam'],autopct="%0.2f")
plt.show()
data is imbalanced spam is less then ham
# for deeper analysis make 3 new column 1.no of character, 2.no of
words, 3.no of sentences so use nltk lib
import nltk

!pip install nltk

Requirement already satisfied: nltk in c:\users\abhishek kumar yadav\

anaconda3\lib\site-packages (3.6.5)
Requirement already satisfied: click in c:\users\abhishek kumar yadav\
anaconda3\lib\site-packages (from nltk) (8.0.3)
Requirement already satisfied: joblib in c:\users\abhishek kumar
yadav\anaconda3\lib\site-packages (from nltk) (1.1.0)
Requirement already satisfied: regex>=2021.8.3 in c:\users\abhishek
kumar yadav\anaconda3\lib\site-packages (from nltk) (2021.8.3)
Requirement already satisfied: tqdm in c:\users\abhishek kumar yadav\
anaconda3\lib\site-packages (from nltk) (4.62.3)
Requirement already satisfied: colorama in c:\users\abhishek kumar
yadav\anaconda3\lib\site-packages (from click->nltk) (0.4.4)

nltk.download('punkt')

[nltk_data] Downloading package punkt to C:\Users\Abhishek Kumar

[nltk_data] Yadav\AppData\Roaming\nltk_data...
[nltk_data] Package punkt is already up-to-date!

True

# count no of character
ds['char_count']=ds['text'].apply(len)

ds.head()
 target text
char_count
0 0 Go until jurong point, crazy.. Available only ...
111
1 0 Ok lar... Joking wif u oni...
29
2 1 Free entry in 2 a wkly comp to win FA Cup fina...
155
3 0 U dun say so early hor... U c already then say...
49
4 0 Nah I don't think he goes to usf, he lives aro...
61

ds['text'].apply(lambda x:nltk.word_tokenize(x))

it divide sms into words

# no of words count, you break sms with there words and count with len
function
words=ds['text'].apply(lambda x:len(nltk.word_tokenize(x)))

# count no of words
ds['word_count']=words

ds.head()

target text
char_count \
0 0 Go until jurong point, crazy.. Available only ...
111
1 0 Ok lar... Joking wif u oni...
29
2 1 Free entry in 2 a wkly comp to win FA Cup fina...
155
3 0 U dun say so early hor... U c already then say...
49
4 0 Nah I don't think he goes to usf, he lives aro...
61

word_count
0 24
1 8
2 37
3 13
4 15

ds['text'].apply(lambda x:nltk.sent_tokenize(x))

it divide sms in terms of sentence

# no of sentences
ds['sent_count']=ds['text'].apply(lambda x:len(nltk.sent_tokenize(x)))
ds.head()

target text
char_count \
0 0 Go until jurong point, crazy.. Available only ...
111
1 0 Ok lar... Joking wif u oni...
29
2 1 Free entry in 2 a wkly comp to win FA Cup fina...
155
3 0 U dun say so early hor... U c already then say...
49
4 0 Nah I don't think he goes to usf, he lives aro...
61

word_count sent_count
0 24 2
1 8 2
2 37 2
3 13 1
4 15 1

# describe all 3 column ,describe as per sms

ds[['char_count','word_count','sent_count']].describe()

char_count word_count sent_count

count 5169.000000 5169.000000 5169.000000
mean 78.977945 18.455407 1.961308
std 58.236293 13.322448 1.432583
min 2.000000 1.000000 1.000000
25% 36.000000 9.000000 1.000000
50% 60.000000 15.000000 1.000000
75% 117.000000 26.000000 2.000000
max 910.000000 220.000000 38.000000

# analysis both spam and ham diffrent , [[]] is list

# this is non spam
ds[ds['target'] ==0 ]
[['char_count','word_count','sent_count']].describe()

char_count word_count sent_count

count 4516.000000 4516.000000 4516.000000
mean 70.459256 17.123339 1.815545
std 56.358207 13.491315 1.364098
min 2.000000 1.000000 1.000000
25% 34.000000 8.000000 1.000000
50% 52.000000 13.000000 1.000000
75% 90.000000 22.000000 2.000000
max 910.000000 220.000000 38.000000
# this is spam
ds[ds['target'] ==1 ]
[['char_count','word_count','sent_count']].describe()

char_count word_count sent_count

count 653.000000 653.000000 653.000000
mean 137.891271 27.667688 2.969372
std 30.137753 7.008418 1.488910
min 13.000000 2.000000 1.000000
25% 132.000000 25.000000 2.000000
50% 149.000000 29.000000 3.000000
75% 157.000000 32.000000 4.000000
max 224.000000 46.000000 9.000000

import seaborn as sns

# histogram form ham(blue) and spam(red)

plt.figure(figsize=(11,6))
sns.histplot(ds[ds['target']==0]['char_count'])
sns.histplot(ds[ds['target']==1]['char_count'],color='red')

<AxesSubplot:xlabel='char_count', ylabel='Count'>

plt.figure(figsize=(11,6))
sns.histplot(ds[ds['target']==0]['word_count'])
sns.histplot(ds[ds['target']==1]['word_count'],color='red')

<AxesSubplot:xlabel='word_count', ylabel='Count'>
# relation between no of column and sentence
sns.pairplot(ds,hue='target')

<seaborn.axisgrid.PairGrid at 0x1d21d92c8e0>
# correlation between table in heatmap
sns.heatmap(ds.corr(),annot=True)

<AxesSubplot:>
Data Preprocessing
.lower case convert .Tokenization (split into words) .Removing special characters(take only
alphabet,aplhanumeric) .Removing stop words and punctuation .Stemming
nltk.download('stopwords')

[nltk_data] Downloading package stopwords to C:\Users\Abhishek Kumar

[nltk_data] Yadav\AppData\Roaming\nltk_data...
[nltk_data] Package stopwords is already up-to-date!

True

import string
from nltk.corpus import stopwords
# for punctuation remove we need string

from nltk.stem.porter import PorterStemmer

ps=PorterStemmer()
ps.stem('loving')

'love'

def transform_text(text):
text=text.lower()
text=nltk.word_tokenize(text)

y=[]
for i in text:
if i.isalnum():
 y.append(i)

text=y[:]
y.clear()

for i in text:
if i not in stopwords.words('english') and i not in
string.punctuation:
y.append(i)

text=y[:]
y.clear()

for i in text:
y.append(ps.stem(i))

return " ".join(y)

transform_text('i loved the yt lecture on mavhine learning. how about

you')

'love yt lectur mavhin learn'

from nltk.corpus import stopwords stopwords.words('english')

import string string.punctuation
you can't copy list not like that text=y value change always
right way of cloning is text = y[:]
transform_text(ds['text'][0])

'go jurong point crazi avail bugi n great world la e buffet cine got
amor wat'

# add another column 'tranform text'

ds['transform_text']=ds['text'].apply(transform_text)

ds.head()

target text
char_count \
0 0 Go until jurong point, crazy.. Available only ...
111
1 0 Ok lar... Joking wif u oni...
29
2 1 Free entry in 2 a wkly comp to win FA Cup fina...
155
3 0 U dun say so early hor... U c already then say...
49
4 0 Nah I don't think he goes to usf, he lives aro...
61

word_count sent_count
transform_text
0 24 2 go jurong point crazi avail bugi n great
world...
1 8 2 ok lar joke wif
u oni
2 37 2 free entri 2 wkli comp win fa cup final tkt
21...
3 13 1 u dun say earli hor u c
alreadi say
4 15 1 nah think goe usf live around
though

conda install -c conda-forge wordcloud=1.6.0

Note: you may need to restart the kernel to use updated packages.

usage: conda-script.py [-h] [-V] command ...

conda-script.py: error: unrecognized arguments: wordcloud=1.6.0

pip install wordcloud

Requirement already satisfied: wordcloud in c:\users\abhishek kumar

yadav\anaconda3\lib\site-packages (1.8.2.2)
Requirement already satisfied: matplotlib in c:\users\abhishek kumar
yadav\anaconda3\lib\site-packages (from wordcloud) (3.4.3)
Requirement already satisfied: pillow in c:\users\abhishek kumar
yadav\anaconda3\lib\site-packages (from wordcloud) (8.4.0)
Requirement already satisfied: numpy>=1.6.1 in c:\users\abhishek kumar
yadav\anaconda3\lib\site-packages (from wordcloud) (1.20.3)
Requirement already satisfied: python-dateutil>=2.7 in c:\users\
abhishek kumar yadav\anaconda3\lib\site-packages (from matplotlib-
>wordcloud) (2.8.2)
Requirement already satisfied: cycler>=0.10 in c:\users\abhishek kumar
yadav\anaconda3\lib\site-packages (from matplotlib->wordcloud)
(0.10.0)
Requirement already satisfied: kiwisolver>=1.0.1 in c:\users\abhishek
kumar yadav\anaconda3\lib\site-packages (from matplotlib->wordcloud)
(1.3.1)
Requirement already satisfied: pyparsing>=2.2.1 in c:\users\abhishek
kumar yadav\anaconda3\lib\site-packages (from matplotlib->wordcloud)
(3.0.4)
Requirement already satisfied: six in c:\users\abhishek kumar yadav\
anaconda3\lib\site-packages (from cycler>=0.10->matplotlib->wordcloud)
(1.16.0)
Note: you may need to restart the kernel to use updated packages.
# lets check what are the most words use in spam or in ham with help
of word cloud
from wordcloud import WordCloud
wc=WordCloud(width=500,height=500,min_font_size=10,background_color='w
hite')

spam_wc=wc.generate(ds[ds['target']==1]
['transform_text'].str.cat(sep=" "))

# spam cloud words

plt.figure(figsize=(15,6))
plt.imshow(spam_wc)

<matplotlib.image.AxesImage at 0x1d21f32e070>

# non spam cloud words

ham_wc=wc.generate(ds[ds['target']==0]['transform_text'].str.cat(sep="
"))
plt.figure(figsize=(15,6))
plt.imshow(ham_wc)

<matplotlib.image.AxesImage at 0x1d221930d30>
# most used 30 word in spam
spam_corpus = []
for msg in ds[ds['target']==1]['transform_text'].tolist():
for word in msg.split():
spam_corpus.append(word)

len(spam_corpus)

9939

from collections import Counter

sns.barplot(pd.DataFrame(Counter(spam_corpus).most_common(20))
[0],pd.DataFrame(Counter(spam_corpus).most_common(20))[1])
plt.xticks(rotation='vertical')
plt.show()

C:\Users\Abhishek Kumar Yadav\anaconda3\lib\site-packages\seaborn\

_decorators.py:36: FutureWarning: Pass the following variables as
keyword args: x, y. From version 0.12, the only valid positional
argument will be `data`, and passing other arguments without an
explicit keyword will result in an error or misinterpretation.
warnings.warn(
# ham most use words so we can diffrenciate between spam or ham
ham_corpus = []
for msg in ds[ds['target']==0]['transform_text'].tolist():
for word in msg.split():
ham_corpus.append(word)

len(ham_corpus)

35402

from collections import Counter

sns.barplot(pd.DataFrame(Counter(ham_corpus).most_common(20))
[0],pd.DataFrame(Counter(ham_corpus).most_common(20))[1])
plt.xticks(rotation='vertical')
plt.show()

C:\Users\Abhishek Kumar Yadav\anaconda3\lib\site-packages\seaborn\

_decorators.py:36: FutureWarning: Pass the following variables as
keyword args: x, y. From version 0.12, the only valid positional
argument will be `data`, and passing other arguments without an
explicit keyword will result in an error or misinterpretation.
warnings.warn(
Model Building
# convert text to number(vector) so machine can predict better

from sklearn.feature_extraction.text import CountVectorizer,

TfidfVectorizer
cv = CountVectorizer()
tfidf=TfidfVectorizer(max_features=3000)

x=tfidf.fit_transform(ds['transform_text']).toarray()

x.shape

(5169, 3000)

y=ds['target'].values

y.shape

(5169,)

from sklearn.model_selection import train_test_split

x_train,x_test,y_train,y_test =
train_test_split(x,y,test_size=0.2,random_state=2)

from sklearn.naive_bayes import GaussianNB,MultinomialNB,BernoulliNB

from sklearn.metrics import
accuracy_score,confusion_matrix,precision_score
gnb=GaussianNB()
mnb=MultinomialNB()
bnb=BernoulliNB()

gnb.fit(x_train,y_train)
y_pred1 = gnb.predict(x_test)
print(accuracy_score(y_test,y_pred1))
print(confusion_matrix(y_test,y_pred1))
print(precision_score(y_test,y_pred1))

0.8694390715667312
[[788 108]
[ 27 111]]
0.5068493150684932

mnb.fit(x_train,y_train)
y_pred2 = mnb.predict(x_test)
print(accuracy_score(y_test,y_pred2))
print(confusion_matrix(y_test,y_pred2))
print(precision_score(y_test,y_pred2))

0.9709864603481625
[[896 0]
[ 30 108]]
1.0

bnb.fit(x_train,y_train)
y_pred3 = bnb.predict(x_test)
print(accuracy_score(y_test,y_pred3))
print(confusion_matrix(y_test,y_pred3))
print(precision_score(y_test,y_pred3))

0.9835589941972921
[[895 1]
[ 16 122]]
0.991869918699187

multinomial nive base is better then other tfidf--> MNB

pip install xgboost

Requirement already satisfied: xgboost in c:\users\abhishek kumar

yadav\anaconda3\lib\site-packages (1.6.2)
Requirement already satisfied: scipy in c:\users\abhishek kumar yadav\
anaconda3\lib\site-packages (from xgboost) (1.7.1)
Requirement already satisfied: numpy in c:\users\abhishek kumar yadav\
anaconda3\lib\site-packages (from xgboost) (1.20.3)
Note: you may need to restart the kernel to use updated packages.

from sklearn.linear_model import LogisticRegression

from sklearn.svm import SVC
from sklearn.naive_bayes import MultinomialNB
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import AdaBoostClassifier
from sklearn.ensemble import BaggingClassifier
from sklearn.ensemble import ExtraTreesClassifier
from sklearn.ensemble import GradientBoostingClassifier
from xgboost import XGBClassifier

svc = SVC(kernel='sigmoid', gamma=1.0)

knc = KNeighborsClassifier()
mnb = MultinomialNB()
dtc = DecisionTreeClassifier(max_depth=5)
lrc = LogisticRegression(solver='liblinear', penalty='l1')
rfc = RandomForestClassifier(n_estimators=50, random_state=2)
abc = AdaBoostClassifier(n_estimators=50, random_state=2)
bc = BaggingClassifier(n_estimators=50, random_state=2)
etc = ExtraTreesClassifier(n_estimators=50, random_state=2)
gbdt = GradientBoostingClassifier(n_estimators=50,random_state=2)
xgb = XGBClassifier(n_estimators=50,random_state=2)

clfs = {
'SVC' : svc,
'KN' : knc,
'NB': mnb,
'DT': dtc,
'LR': lrc,
'RF': rfc,
'AdaBoost': abc,
'BgC': bc,
'ETC': etc,
'GBDT':gbdt,
'xgb':xgb
}

def train_classifier(clf,X_train,y_train,X_test,y_test):
clf.fit(X_train,y_train)
y_pred = clf.predict(X_test)
accuracy = accuracy_score(y_test,y_pred)
precision = precision_score(y_test,y_pred)

return accuracy,precision

train_classifier(svc,x_train,y_train,x_test,y_test)

(0.9758220502901354, 0.9747899159663865)

accuracy_scores = []
precision_scores = []

for name,clf in clfs.items():

 current_accuracy,current_precision = train_classifier(clf,
x_train,y_train,x_test,y_test)

print("For ",name)
print("Accuracy - ",current_accuracy)
print("Precision - ",current_precision)

accuracy_scores.append(current_accuracy)
precision_scores.append(current_precision)

For SVC
Accuracy - 0.9758220502901354
Precision - 0.9747899159663865
For KN
Accuracy - 0.9052224371373307
Precision - 1.0
For NB
Accuracy - 0.9709864603481625
Precision - 1.0
For DT
Accuracy - 0.9274661508704062
Precision - 0.8118811881188119
For LR
Accuracy - 0.9584139264990329
Precision - 0.9702970297029703
For RF
Accuracy - 0.9758220502901354
Precision - 0.9829059829059829
For AdaBoost
Accuracy - 0.960348162475822
Precision - 0.9292035398230089
For BgC
Accuracy - 0.9584139264990329
Precision - 0.8682170542635659
For ETC
Accuracy - 0.9748549323017408
Precision - 0.9745762711864406
For GBDT
Accuracy - 0.9468085106382979
Precision - 0.9191919191919192
For xgb
Accuracy - 0.9671179883945842
Precision - 0.9333333333333333

performance_df =
pd.DataFrame({'Algorithm':clfs.keys(),'Accuracy':accuracy_scores,'Prec
ision':precision_scores}).sort_values('Precision',ascending=False)

performance_df
 Algorithm Accuracy Precision
1 KN 0.905222 1.000000
2 NB 0.970986 1.000000
5 RF 0.975822 0.982906
0 SVC 0.975822 0.974790
8 ETC 0.974855 0.974576
4 LR 0.958414 0.970297
10 xgb 0.967118 0.933333
6 AdaBoost 0.960348 0.929204
9 GBDT 0.946809 0.919192
7 BgC 0.958414 0.868217
3 DT 0.927466 0.811881

performance_df1 = pd.melt(performance_df, id_vars = "Algorithm")

performance_df1

Algorithm variable value

0 KN Accuracy 0.905222
1 NB Accuracy 0.970986
2 RF Accuracy 0.975822
3 SVC Accuracy 0.975822
4 ETC Accuracy 0.974855
5 LR Accuracy 0.958414
6 xgb Accuracy 0.967118
7 AdaBoost Accuracy 0.960348
8 GBDT Accuracy 0.946809
9 BgC Accuracy 0.958414
10 DT Accuracy 0.927466
11 KN Precision 1.000000
12 NB Precision 1.000000
13 RF Precision 0.982906
14 SVC Precision 0.974790
15 ETC Precision 0.974576
16 LR Precision 0.970297
17 xgb Precision 0.933333
18 AdaBoost Precision 0.929204
19 GBDT Precision 0.919192
20 BgC Precision 0.868217
21 DT Precision 0.811881

sns.catplot(x = 'Algorithm', y='value',

hue = 'variable',data=performance_df1,
kind='bar',height=5)
plt.ylim(0.5,1.0)
plt.xticks(rotation='vertical')
plt.show()
# model improvement
# 1. change maximum_feature parameter of Tfidf

temp_df =
pd.DataFrame({'Algorithm':clfs.keys(),'Accuracy_max_ft_3000':accuracy_
scores,'Precision_max_ft_3000':precision_scores}).sort_values('Precisi
on_max_ft_3000',ascending=False)

temp_df =
pd.DataFrame({'Algorithm':clfs.keys(),'Accuracy_scaling':accuracy_scor
es,'Precision_scaling':precision_scores}).sort_values('Precision_scali
ng',ascending=False)

new_df = performance_df.merge(temp_df,on='Algorithm')

new_df_scaled = new_df.merge(temp_df,on='Algorithm')

temp_df =
pd.DataFrame({'Algorithm':clfs.keys(),'Accuracy_num_chars':accuracy_sc
ores,'Precision_num_chars':precision_scores}).sort_values('Precision_n
um_chars',ascending=False)

new_df_scaled.merge(temp_df,on='Algorithm')

Algorithm Accuracy Precision Accuracy_max_ft_3000 \

0 KN 0.905222 1.000000 0.905222
1 NB 0.970986 1.000000 0.970986
2 RF 0.975822 0.982906 0.975822
3 SVC 0.975822 0.974790 0.975822
4 ETC 0.974855 0.974576 0.974855
5 LR 0.958414 0.970297 0.958414
6 xgb 0.967118 0.933333 0.967118
7 AdaBoost 0.960348 0.929204 0.960348
8 GBDT 0.946809 0.919192 0.946809
9 BgC 0.958414 0.868217 0.958414
10 DT 0.927466 0.811881 0.927466

Precision_max_ft_3000 Accuracy_scaling Precision_scaling \

0 1.000000 0.905222 1.000000
1 1.000000 0.970986 1.000000
2 0.982906 0.975822 0.982906
3 0.974790 0.975822 0.974790
4 0.974576 0.974855 0.974576
5 0.970297 0.958414 0.970297
6 0.933333 0.967118 0.933333
7 0.929204 0.960348 0.929204
8 0.919192 0.946809 0.919192
9 0.868217 0.958414 0.868217
10 0.811881 0.927466 0.811881

Accuracy_num_chars Precision_num_chars
0 0.905222 1.000000
1 0.970986 1.000000
2 0.975822 0.982906
3 0.975822 0.974790
4 0.974855 0.974576
5 0.958414 0.970297
6 0.967118 0.933333
7 0.960348 0.929204
8 0.946809 0.919192
9 0.958414 0.868217
10 0.927466 0.811881

# Voting Classifier
svc = SVC(kernel='sigmoid', gamma=1.0,probability=True)
mnb = MultinomialNB()
etc = ExtraTreesClassifier(n_estimators=50, random_state=2)

from sklearn.ensemble import VotingClassifier

voting = VotingClassifier(estimators=[('svm', svc), ('nb', mnb),
('et', etc)],voting='soft')

voting.fit(x_train,y_train)

VotingClassifier(estimators=[('svm',
SVC(gamma=1.0, kernel='sigmoid',
probability=True)),
('nb', MultinomialNB()),
('et',
ExtraTreesClassifier(n_estimators=50,
random_state=2))],
voting='soft')

y_pred = voting.predict(x_test)
print("Accuracy",accuracy_score(y_test,y_pred))
print("Precision",precision_score(y_test,y_pred))

Accuracy 0.9816247582205029
Precision 0.9917355371900827

# Applying stacking
estimators=[('svm', svc), ('nb', mnb), ('et', etc)]
final_estimator=RandomForestClassifier()

from sklearn.ensemble import StackingClassifier

clf = StackingClassifier(estimators=estimators,
final_estimator=final_estimator)

clf.fit(x_train,y_train)
y_pred = clf.predict(x_test)
print("Accuracy",accuracy_score(y_test,y_pred))
print("Precision",precision_score(y_test,y_pred))

Accuracy 0.9806576402321083
Precision 0.9538461538461539

import pickle
pickle.dump(tfidf,open('vectorizer.pkl','wb'))
pickle.dump(mnb,open('model.pkl','wb'))