13/09/2023, 21:06 principal-component-analysis
Import all the libraries :
In [22]:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
Loading Data :
In [23]:
df = pd.read_csv('/kaggle/input/principal-component-analysis/Longley (1).csv')
In [24]:
df
Out[24]:
GNP.deflator GNP Unemployed Armed.Forces Population Employed
0 83.0 234.289 235.6 159.0 107.608 60.323
1 88.5 259.426 232.5 145.6 108.632 61.122
2 88.2 258.054 368.2 161.6 109.773 60.171
3 89.5 284.599 335.1 165.0 110.929 61.187
4 96.2 328.975 209.9 309.9 112.075 63.221
5 98.1 346.999 193.2 359.4 113.270 63.639
6 99.0 365.385 187.0 354.7 115.094 64.989
7 100.0 363.112 357.8 335.0 116.219 63.761
8 101.2 397.469 290.4 304.8 117.388 66.019
9 104.6 419.180 282.2 285.7 118.734 67.857
10 108.4 442.769 293.6 279.8 120.445 68.169
11 110.8 444.546 468.1 263.7 121.950 66.513
12 112.6 482.704 381.3 255.2 123.366 68.655
13 114.2 502.601 393.1 251.4 125.368 69.564
14 115.7 518.173 480.6 257.2 127.852 69.331
15 116.9 554.894 400.7 282.7 130.081 70.551
https://htmtopdf.herokuapp.com/ipynbviewer/temp/4518ed9eb5f5a3af5f67858dbb1814e4/principal-component-analysis.html?t=1694619339010 1/4
13/09/2023, 21:06 principal-component-analysis
In [25]:
df.dtypes
Out[25]:
GNP.deflator float64
GNP float64
Unemployed float64
Armed.Forces float64
Population float64
Employed float64
dtype: object
In [26]:
X = df.drop('Employed', axis=1)
Y = df['Employed']
In [27]:
correlation = df.corr()
correlation
Out[27]:
GNP.deflator GNP Unemployed Armed.Forces Population Employed
GNP.deflator 1.000000 0.991589 0.620633 0.464744 0.979163 0.970899
GNP 0.991589 1.000000 0.604261 0.446437 0.991090 0.983552
Unemployed 0.620633 0.604261 1.000000 -0.177421 0.686552 0.502498
Armed.Forces 0.464744 0.446437 -0.177421 1.000000 0.364416 0.457307
Population 0.979163 0.991090 0.686552 0.364416 1.000000 0.960391
Employed 0.970899 0.983552 0.502498 0.457307 0.960391 1.000000
Apply PCA :
In [28]:
from sklearn.preprocessing import StandardScaler
In [29]:
# Scale data before applying PCA
scaling=StandardScaler()
https://htmtopdf.herokuapp.com/ipynbviewer/temp/4518ed9eb5f5a3af5f67858dbb1814e4/principal-component-analysis.html?t=1694619339010 2/4
13/09/2023, 21:06 principal-component-analysis
In [30]:
# Use fit and transform method
scaling.fit(df)
Scaled_data=scaling.transform(df)
In [31]:
from sklearn.decomposition import PCA
In [32]:
# Set the n_components=3
principal=PCA(n_components=3)
principal.fit(Scaled_data)
x=principal.transform(Scaled_data)
In [33]:
# Check the dimensions of data after PCA
print(x.shape)
(16, 3)
Check Components :
In [34]:
# Check the values of eigen vectors
# prodeced by principal components
principal.components_
Out[34]:
array([[-0.46695493, -0.46748987, -0.30646472, -0.21200613, -0.4656055
6,
-0.45579661],
[ 0.02628724, 0.02306569, -0.62227098, 0.77353962, -0.0762474
5,
0.08589854],
[-0.04906877, -0.16405382, 0.67228378, 0.58400807, -0.0917922
6,
-0.41136586]])
Plot the components (Visualization) :
https://htmtopdf.herokuapp.com/ipynbviewer/temp/4518ed9eb5f5a3af5f67858dbb1814e4/principal-component-analysis.html?t=1694619339010 3/4
13/09/2023, 21:06 principal-component-analysis
In [35]:
# plt.figure(figsize=(10,10))
plt.scatter(x[:,0],x[:,1],c=df['Employed'],cmap='plasma')
plt.xlabel('pc1')
plt.ylabel('pc2')
Out[35]:
Text(0, 0.5, 'pc2')
Calculate variance ratio :
In [36]:
# check how much variance is explained by each principal component
print(principal.explained_variance_ratio_)
[0.75584735 0.19778211 0.0419845 ]
In [ ]:
https://htmtopdf.herokuapp.com/ipynbviewer/temp/4518ed9eb5f5a3af5f67858dbb1814e4/principal-component-analysis.html?t=1694619339010 4/4
