Question

Use Support Vector Machines to solve a text-mining problem of your choice.

• - Define your text mining dataset and problem

• - Run SVM in Python, R or Weka to gain knowledge from the dataset

• - Generate results, screenshots of any interesting observations along the way

• - Discuss the results

Answer 1

#This peice of code is to solve a classification problem on a well known iris dataset using SVM in python
#This iris dataset comes with python package 'sklearn'
#in this data set we have 5 variables in total
# four predictors or independent variables(sepal_length, sepal_width', 'petal_length', 'petal_width') which
#are continuous in nature and will help us classify the species(the target variable)
# species variables takes three possible values as setosa, versicolor and virginica
# Our goal is run SVM classifier to predict species on test data
# we split the whole dataset into training and testing datasets into 80:20 ratio
# Run SVM on training data and run the classifier on test data

#importing python packages used for ML
import numpy as np
import pandas as pd
from sklearn import svm
from sklearn import datasets

#loading iris datase
#Iris dataset has four predictors as sepal_length, sepal_width', 'petal_length', 'petal_width' and target variable 'species'
iris = datasets.load_iris()

# fetching predictors and target variables out dataset
# y has three possible values 0,1 and 2 where
# 0 --> setosa
# 1 --> versicolor
# 2 --> virginica

X, y = iris.data, iris.target

#splitting data into training and test set
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.20)

#defining SVM classifier
clf = svm.SVC(gamma='scale')

#fitting model to the training data
clf.fit(X_train, y_train)

#Now making prections on testing dataset using model we have achieved

y_pred = clf.predict(X_test)
print(y_pred)

#evaluating model performance
from sklearn.metrics import accuracy_score
accuracy = accuracy_score(y_test, y_pred)

print(accuracy)

#As we can see model performa well as its accuracy is 94%, which is very good.