[PYTHON] <Course> Machine learning Chapter 4: Principal component analysis

Machine learning

table of contents Chapter 1: Linear Regression Model [Chapter 2: Nonlinear Regression Model] (https://qiita.com/matsukura04583/items/baa3f2269537036abc57) [Chapter 3: Logistic Regression Model] (https://qiita.com/matsukura04583/items/0fb73183e4a7a6f06aa5) [Chapter 4: Principal Component Analysis] (https://qiita.com/matsukura04583/items/b3b5d2d22189afc9c81c) [Chapter 5: Algorithm 1 (k-nearest neighbor method (kNN))] (https://qiita.com/matsukura04583/items/543719b44159322221ed) [Chapter 6: Algorithm 2 (k-means)] (https://qiita.com/matsukura04583/items/050c98c7bb1c9e91be71) [Chapter 7: Support Vector Machine] (https://qiita.com/matsukura04583/items/6b718642bcbf97ae2ca8)

Chapter 4: Principal Component Analysis

What is principal component analysis?

• Compress the structure of multivariate data into a smaller number of indicators
• I want to minimize the loss of information associated with reducing the number of variables
• Achieves analysis and visualization (in the case of 2D and 3D) using decimal variables

• If the coefficient vector changes, the value after linear transformation changes
• Consider the amount of information as the size of the variance
• Search for the projection axis that maximizes the variance of the variables after linear transformation 90 Variance after linear transformation 90 Principal component analysis

• Solve the following constrained optimization problem
• Insert a constraint whose norm is 1 (without the constraint, there are infinite solutions)

(Practice 4) Dimensional compression of breast cancer test data using scikit learn

• Settings
• Create a logistic regression model using breast cancer test data
• Dimensional compression in 2D space using principal components
• Number of records 569 Number of columns 33
• Challenges
• Check if it can be discriminated well when 32D data is compressed 2D.

Start with Google Drive Maps

from google.colab import drive
drive.mount('/content/drive')

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegressionCV
from sklearn.metrics import confusion_matrix
from sklearn.decomposition import PCA
import matplotlib.pyplot as plt
%matplotlib inline

sys.path settings

Below, we are creating a study_ai_ml folder directly under My Drive in Google Drive.

cancer_df = pd.read_csv('/content/drive/My Drive/study_ai_ml/data/cancer.csv')
print('cancer df shape: {}'.format(cancer_df.shape))

result

cancer df shape: (569, 33)

cancer_df

cancer_df.drop('Unnamed: 32', axis=1, inplace=True)
cancer_df

** ・ Diagnosis: Diagnosis result (Benign is B / Malignant is M) ・ Explanatory variables are classified by logistic regression with the objective variable in the second column after the third column **

#Extraction of objective variable
y = cancer_df.diagnosis.apply(lambda d: 1 if d == 'M' else 0)
#Extraction of explanatory variables
X = cancer_df.loc[:, 'radius_mean':]
#Separate data for learning and testing
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)

#Standardization
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

#Learning with logistic regression
logistic = LogisticRegressionCV(cv=10, random_state=0)
logistic.fit(X_train_scaled, y_train)

#Verification
print('Train score: {:.3f}'.format(logistic.score(X_train_scaled, y_train)))
print('Test score: {:.3f}'.format(logistic.score(X_test_scaled, y_test)))
print('Confustion matrix:\n{}'.format(confusion_matrix(y_true=y_test, y_pred=logistic.predict(X_test_scaled))))

result

Train score: 0.988
Test score: 0.972
Confustion matrix:
[[89  1]
 [ 3 50]]

** ・ Confirmed that it can be classified with a verification score of 97% **

pca = PCA(n_components=30)
pca.fit(X_train_scaled)
plt.bar([n for n in range(1, len(pca.explained_variance_ratio_)+1)], pca.explained_variance_ratio_)

# PCA
#Compress up to 2 dimensions
pca = PCA(n_components=2)
X_train_pca = pca.fit_transform(X_train_scaled)
print('X_train_pca shape: {}'.format(X_train_pca.shape))
# X_train_pca shape: (426, 2)

#Contribution rate
print('explained variance ratio: {}'.format(pca.explained_variance_ratio_))
# explained variance ratio: [ 0.43315126  0.19586506]

#Plot on scatter plot
temp = pd.DataFrame(X_train_pca)
temp['Outcome'] = y_train.values
b = temp[temp['Outcome'] == 0]
m = temp[temp['Outcome'] == 1]
plt.scatter(x=b[0], y=b[1], marker='o') #Benign is marked with ○
plt.scatter(x=m[0], y=m[1], marker='^') #Malignant is marked with △
plt.xlabel('PC 1') #X-axis of the first principal component
plt.ylabel('PC 2') #The second principal component is the y-axis

#Standardization
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

#Learning with logistic regression
logistic = LogisticRegressionCV(cv=10, random_state=0)
logistic.fit(X_train_scaled, y_train)

#Verification
print('Train score: {:.3f}'.format(logistic.score(X_train_scaled, y_train)))
print('Test score: {:.3f}'.format(logistic.score(X_test_scaled, y_test)))
print('Confustion matrix:\n{}'.format(confusion_matrix(y_true=y_test, y_pred=logistic.predict(X_test_scaled))))

result

Train score: 0.927
Test score: 0.944
Confustion matrix:
[[87  3]
 [ 5 48]]

** ・ Confirmed that it can be classified with a verification score of 94% ** Even if the number of dimensions was reduced to 2, the verification score did not drop much from 97% to 94%, and the result was that the number of dimensions was reduced while maintaining accuracy.

Related Sites

Chapter 1: Linear Regression Model [Chapter 2: Nonlinear Regression Model] (https://qiita.com/matsukura04583/items/baa3f2269537036abc57) [Chapter 3: Logistic Regression Model] (https://qiita.com/matsukura04583/items/0fb73183e4a7a6f06aa5) [Chapter 4: Principal Component Analysis] (https://qiita.com/matsukura04583/items/b3b5d2d22189afc9c81c) [Chapter 5: Algorithm 1 (k-nearest neighbor method (kNN))] (https://qiita.com/matsukura04583/items/543719b44159322221ed) [Chapter 6: Algorithm 2 (k-means)] (https://qiita.com/matsukura04583/items/050c98c7bb1c9e91be71) [Chapter 7: Support Vector Machine] (https://qiita.com/matsukura04583/items/6b718642bcbf97ae2ca8)