20BECE30058

1) Preprocessing

Fetching the Dataset

import numpy as np
import pandas as pd

dataset=pd.read_csv('/content/drive/MyDrive/ML/ML_Lab/Data.csv')

Dataset

Country Age Salary Purchased

0 France 44.0 72000.0 No

1 Spain 27.0 48000.0 Yes

2 Germany NaN 54000.0 No

3 Spain 38.0 61000.0 No

4 Germany 40.0 NaN Yes

5 France 35.0 58000.0 Yes

6 Spain NaN 52000.0 No

7 France 48.0 79000.0 Yes

8 Germany 50.0 NaN No

9 France 37.0 67000.0 Yes

from google.colab import drive

drive.mount('/content/drive')

Mounted at /content/drive

X=dataset.iloc[:,:-1].values #---taking all the columns except the label

column y=dataset.iloc[:,-1].values #---taking the values of the label column
print(X) print(y)

[['France' 44.0 72000.0]

['Spain' 27.0 48000.0]
['Germany' nan 54000.0]
['Spain' 38.0 61000.0]
['Germany' 40.0 nan]
['France' 35.0 58000.0]
['Spain' nan 52000.0]
['France' 48.0 79000.0]
['Germany' 50.0 nan]
['France' 37.0 67000.0]]
['No' 'Yes' 'No' 'No' 'Yes' 'Yes' 'No' 'Yes' 'No' 'Yes']

Handling missing data

from sklearn.impute import SimpleImputer

#---create an instance of the class SimpleImputer imputer=SimpleImputer(missing_values=np.nan, strategy= 'mean') #---

missing_values=np.nan, means all the empty values in the data
#---strategy=mean, median, most_frequent, constant

#---connect the imputer object with the data imputer.fit(X[:,1:3]) #---the column with
numerical data is only fitted to imputer object

#----transform the data(replace the missing data with the mean)

X[:,1:3]=imputer.transform(X[:,1:3]) print(X) #------no missing

values now in the X

[['France' 44.0 72000.0]

['Spain' 27.0 48000.0]
['Germany' 39.875 54000.0]
['Spain' 38.0 61000.0]
['Germany' 40.0 61375.0]
['France' 35.0 58000.0]
['Spain' 39.875 52000.0]
['France' 48.0 79000.0]

https://colab.research.google.com/drive/1epQs6N97ffLdHEAPXVRF8eYmSDkR565_#scrollTo=_IsOM8AA5_QF&printMode=true 1
 20BECE30058
['Germany' 50.0 61375.0]
['France' 37.0 67000.0]]

----Arguments----

SimpleImputer( missing_values=nan, strategy='mean', ll_value=None, verbose=0, copy=True, add_indicator=False, ) copy-->If

True, a copy of X will be created. If False, imputation will be done in-place whenever possible. ll_value--> When strategy ==

"constant", ll_value is used to replace all occurrences of missing_values.default=None

Encoding Categorical data

Suppose the data is categorical in the dataset and we need to convert it into the numeric format

Here we are having "Purchased" & "Country" column as categorical data

'Country' has 3 values--> France, Spain, Germany, If we assign 0, 1 and 2 numbers respectively to each value then it will be considered as some kind of
priority given to each value(0,1 & 2)

To do this, we can do one hot encoding(create different columns for each unique values)

from sklearn.compose import ColumnTransformer

from sklearn.preprocessing import OneHotEncoder

#-----Object of Column Transformer Class ct= ColumnTransformer(transformers=

[('encoder',OneHotEncoder(),[0])], remainder='passthrough')
#---3 things in 'transformers'= 1st--> kind of
transformation/encoding #--- 2nd--> Class of encoder
#--- 3rd--> index of the columns that we want to apply the OnHot encoding,here [0] for country column #-
--remainder=passthrough --> that means we want to keep all other features/columns in feature matrix without
transformation # {'drop','passthrough'}

print(type(X))
#-----Connect the data with the 'ct' object,here we will use 'fit_transform' which will fit and transform together in 1 step
X=ct.fit_transform(X) #-- - -update the encoded column to the same feature matrix X
X

<class 'numpy.ndarray'>
array([[1.0, 0.0, 0.0, 44.0, 72000.0],
[0.0, 0.0, 1.0, 27.0, 48000.0],
[0.0, 1.0, 0.0, 39.875, 54000.0],
[0.0, 0.0, 1.0, 38.0, 61000.0],
[0.0, 1.0, 0.0, 40.0, 61375.0],
[1.0, 0.0, 0.0, 35.0, 58000.0],
[0.0, 0.0, 1.0, 39.875, 52000.0],
[1.0, 0.0, 0.0, 48.0, 79000.0],
[0.0, 1.0, 0.0, 50.0, 61375.0],
[1.0, 0.0, 0.0, 37.0, 67000.0]], dtype=object)

Encoding the dependent Variable

from sklearn.preprocessing import LabelEncoder

le=LabelEncoder() y=le.fit_transform(y) #---parameter as

dependent variable print(y)

[0 1 0 0 1 1 0 1 0 1]

Splitting Dataset
from sklearn.model_selection import train_test_split
#---returns the 4 matrices-2 for training and 2 for testing

X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.2,random_state=1) #--random_state=seed value for splitting order #-

--X stands for feature matrix and y stands for target matrix

print(X_train)

https://colab.research.google.com/drive/1epQs6N97ffLdHEAPXVRF8eYmSDkR565_#scrollTo=_IsOM8AA5_QF&printMode=true 2
 20BECE30058
[[0.0 0.0 1.0 39.875 52000.0]
[0.0 1.0 0.0 40.0 61375.0]
[1.0 0.0 0.0 44.0 72000.0]
[0.0 0.0 1.0 38.0 61000.0]
[0.0 0.0 1.0 27.0 48000.0]
[1.0 0.0 0.0 48.0 79000.0]
[0.0 1.0 0.0 50.0 61375.0]
[1.0 0.0 0.0 35.0 58000.0]]

print(X_test)

[[0.0 1.0 0.0 39.875 54000.0]

[1.0 0.0 0.0 37.0 67000.0]]

print(y_train)

[0 1 0 0 1 1 0 1]

print(y_test)

[0 1]

Feature Scaling

Feature scaling is done to prevent over dominate some feature values just based upon their magnitues even if they were not important

2 Methods of Feature Scaling

Standardization (scales between -3 and +3)

x(stand)=[x-mean(x)]/standard deviation(x)

Normalization (scales between 0 and +1) x

(norm)=[x-min(x)]/max(x)-min(x)

Normalization is recomended when you have a normal distribution in features Standardization is

used in general conditions

from sklearn.preprocessing import StandardScaler

sc=StandardScaler()
X_train[:,3:]=sc.fit_transform(X_train[:,3:]) #---specify the range, takes all rows and selected columns from 3 upto last column
#---fit method will only compute the mean and standard deviation of the feature,then transform method will apply the

formula # we apply the same scaler to tes set, thats why we use 'transform' method only

X_test[:,3:]=sc.transform(X_test[:,3:])

print(X_train)

[[0.0 0.0 1.0 -0.05231070414657415 -1.0245464314521104]

[0.0 1.0 0.0 -0.03411567661733096 -0.023360993551025323]
[1.0 0.0 0.0 0.5481252043184508 1.1113158360702047]
[0.0 0.0 1.0 -0.3252361170852219 -0.06340841106706872]
[0.0 0.0 1.0 -1.9263985396586218 -1.4517188849565736]
[1.0 0.0 0.0 1.1303660852542325 1.858867629703015]
[0.0 1.0 0.0 1.4214865257221234 -0.023360993551025323]
[1.0 0.0 0.0 -0.7619167777870582 -0.38378775119541597]]

print(X_test)

[[0.0 1.0 0.0 -0.05231070414657415 -0.8109602046998791]

[1.0 0.0 0.0 -0.4707963373191673 0.5773502691896258]]

Note : We need to apply the same scaler of training data on the test data, otherwise it will compute the different values of mean and standard
deviation for the test set and the results would be affected. So that we will get the same transformation on the test set

Do we have to apply the scaling to the dummy variables/ one hot encoders? No, as the scaling takes the values between -3 and +3, the values in dummy
varibales lie between 0 and 1 which fall in this range.

Apply feature scaling to numerical values only, not to dummy variables.

https://colab.research.google.com/drive/1epQs6N97ffLdHEAPXVRF8eYmSDkR565_#scrollTo=_IsOM8AA5_QF&printMode=true 3
 20BECE30058

https://colab.research.google.com/drive/1epQs6N97ffLdHEAPXVRF8eYmSDkR565_#scrollTo=_IsOM8AA5_QF&printMode=true 4