In
[ ]: # SVM when output variable have discrete column
In [2]: import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
In [3]: iris = pd.read_csv("D:\\Course\\Python\\Datasets\\iris.csv")
iris
...
In [4]: iris_setosa = iris.iloc[0:50,0:5] # 50
iris_versicolor = iris.iloc[50:100,0:5] # 50
iris_vergenica = iris.iloc[100:150,0:5] # 50
In [5]: iris_versicolor
...
In [7]: # first 25 observation from setosa data - df1
df1 = pd.DataFrame(iris_setosa.iloc[0:25,0:5])
In [8]: # first 25 obesrvation from Versicolor data - df2
df2 = pd.DataFrame(iris_versicolor.iloc[0:25,0:5])
#df2
In [9]: # first 25 observations from Vergenica dataset - df3
df3 = pd.DataFrame(iris_vergenica.iloc[0:25,0:5])
In [10]: # training dataset which contains 25 from all 3 datasets by using concat
iris_train = pd.concat([df1,df2,df3])
In [11]: iris_train
...
In [12]: # Dividing the input and output variable for Training Dataset - xtrain and Y trai
X_train = iris_train.iloc[:,0:4]
X_train
y_train = iris_train.iloc[:,4]
y_train
X_train
...
In [13]: # next 25 observation from Setosa dataset
df1 = pd.DataFrame(iris_setosa.iloc[25:50,0:5])
# next 25 observation from Versicolor dataset
df2 = pd.DataFrame(iris_versicolor.iloc[25:50,0:5])
# next 25 observation from Vergenica dataset
df3 = pd.DataFrame(iris_vergenica.iloc[25:50,0:5])
In [14]: # testing datset by concat the all 3 datasets ( df1,df2,df3)
iris_test= pd.concat([df1,df2,df3])
iris_test
...
In [15]: iris_test.shape
Out[15]: (75, 5)
In [16]: X_test = iris_test.iloc[:,0:4]
X_test
y_test = iris_test.iloc[:,4]
y_test
...
In [26]: # X_train and X_test are the dataset for training and Testing
#y_train ,y_test are target predictor for Training and Testing Dataset
In [17]: X_train, y_train,X_test,y_test
...
In [18]: # Fitting SVM to the Training set
# we dont have radial Method in python so we have to use linear method for linear
# Importing libraries and Method
from sklearn.svm import SVC
# Creating An alogrithm using imported method names
classifier = SVC(kernel = 'linear')
#Apply alogrithm on Training dataset (xtrain , ytrain) by using .fitI()
classifier.fit(X_train, y_train)
...
In [19]: # Predicting the Test set results by applying on only input variables of
# testind dataset (Xtest)
y_pred = classifier.predict(X_test)
In [20]: y_pred
...
In [22]: y_test
...
In [23]: # Making the Confusion Matrix by validating the predicted and Actual Values
# importing required libraries
from sklearn.metrics import confusion_matrix
cm = confusion_matrix(y_test, y_pred)
In [24]: cm
Out[24]: array([[25, 0, 0],
[ 0, 24, 1],
[ 0, 3, 22]], dtype=int64)
In [25]: from sklearn.metrics import accuracy_score
Accuracy_Score = accuracy_score(y_test, y_pred)
In [26]: Accuracy_Score
Out[26]: 0.9466666666666667
In [29]: # WORKING ON KERNET TRICK
In [27]: # Fitting SVM to the Training set
# we dont have radial Method in python so we have to use linear method for linear
from sklearn.svm import SVC
classifier1 = SVC(kernel = 'poly')
classifier1.fit(X_train, y_train)
...
In [32]: # Predicting the Test set results
y_pred1 = classifier1.predict(X_test)
y_pred1
...
In [33]: # Making the Confusion Matrix
from sklearn.metrics import confusion_matrix
cm = confusion_matrix(y_test, y_pred1)
cm
...
In [34]: from sklearn.metrics import accuracy_score
Accuracy_Score = accuracy_score(y_test, y_pred1)
In [35]: Accuracy_Score
Out[35]: 0.9333333333333333
In [33]: # Kernel Trick for non linear graph
# 1)polynomial - poly
# 2) Gaussian - rbf
# 3) Sigmoid - sigmoid
