What I learned about AI / machine learning using Python (3)

Introduction

I'm studying with this book How to make AI / machine learning / deep learning apps with Python

scikit-learn A classic machine learning framework for Python http://scikit-learn.org/

It has the following features

--Supports various algorithms used in machine learning -Sample data is included so that you can try machine learning immediately --Has a function to verify the results of machine learning --High affinity with other libraries often used in machine learning (Pnadas, Numpy, scipy, Matplotlib etc) --Free commercial use due to BSD licensed open source

Select an algorithm

https://scikit-learn.org/stable/tutorial/machine_learning_map/index.html You can select an algorithm by following the conditions such as what kind of machine learning you want to do and what kind of data you are preparing.

Machine learning AND operation

and.py

#Import the library
from sklearn.svm import LinearSVC            #Package for utilizing the LinearSVC algorithm(sklearn.svm.LinearSVC)
from sklearn.metrics import accuracy_score   #Package for evaluating test results(sklearn.metrics.accuracy_score)
                           

#Prepare learning data
learn_data  = [[0, 0], [1, 0], [0, 1], [1, 1]] #Input data for learning(AND input value)
learn_label = [0, 0, 0, 1]                     #Result data for learning(AND output value)

#Algorithm specification
clf = LinearSVC()  

#learn(Pass input data and result data for learning)
clf.fit(learn_data, learn_label)               #Pass input data and result data for learning

#Verify learning
test_data = [[0, 0], [1, 0], [0, 1], [1, 1]]   #Prepare input data for test data
test_label = clf.predict(test_data)            #Get test data results

#Display test data results
print(test_data, "Prediction result", test_label)

#Display correct answer rate
Accuracy_rate= accuracy_score([0, 0, 0, 1], test_label) # accuracy_score(Correct answer data,Test results)
print("Correct answer rate:", Accuracy_rate)

[[0, 0], [1, 0], [0, 1], [1, 1]]Prediction result[0 0 0 1]
Correct answer rate: 1.0

You are learning AND logical operations correctly. Those that can be classified linearly can be solved by the LinearSVC algorithm. But what about XOR?

Words used

• SVM(Support Vector Machine) "Some data that is really necessary for forecasting" This site is easy to understand

--Linear SVM Classification ... Linear SVM Classification --accuracy ... accuracy --classification ... classification --classifier ... classifier --fit ... learn (fit) --predict ... predict

Machine learning XOR operation

xor.py

#Import the library
from sklearn.svm import LinearSVC            #Package for utilizing the LinearSVC algorithm(sklearn.svm.LinearSVC)
from sklearn.metrics import accuracy_score   #Package for evaluating test results(sklearn.metrics.accuracy_score)
                           

#Prepare learning data
learn_data  = [[0, 0], [1, 0], [0, 1], [1, 1]] #Input data for learning(XOR input value)
learn_label = [0, 1, 1, 0]                     #Result data for learning(XOR output value)

#Algorithm specification
clf = LinearSVC()  

#learn(Pass input data and result data for learning)
clf.fit(learn_data, learn_label)               #Pass input data and result data for learning

#Verify learning
test_data = [[0, 0], [1, 0], [0, 1], [1, 1]]   #Prepare input data for test data
test_label = clf.predict(test_data)            #Get test data results

#Display test data results
print(test_data, "Prediction result", test_label)

#Display correct answer rate
Accuracy_rate= accuracy_score([0, 1, 1, 0], test_label) # accuracy_score(Correct answer data,Test results)
print("Correct answer rate:", Accuracy_rate)

[[0, 0], [1, 0], [0, 1], [1, 1]]Prediction result[0 0 0 0]
Correct answer rate: 0.5

Well, it can't be classified correctly by linear.

xor2.py

#Import the library
from sklearn.neighbors import KNeighborsClassifier #Package for utilizing the LinearSVC algorithm(sklearn.svm.LinearSVC)
from sklearn.metrics import accuracy_score         #Package for evaluating test results(sklearn.metrics.accuracy_score)


#Prepare learning data
learn_data  = [[0, 0], [1, 0], [0, 1], [1, 1]] #Input data for learning(XOR input value)
learn_label = [0, 1, 1, 0]                     #Result data for learning(XOR output value)

#Algorithm specification
clf = KNeighborsClassifier(n_neighbors = 1)  

#learn(Pass input data and result data for learning)
clf.fit(learn_data, learn_label)               #Pass input data and result data for learning

#Verify learning
test_data = [[0, 0], [1, 0], [0, 1], [1, 1]]   #Prepare input data for test data
test_label = clf.predict(test_data)            #Get test data results

#Display test data results
print(test_data, "Prediction result", test_label)

#Display correct answer rate
Accuracy_rate= accuracy_score([0, 1, 1, 0], test_label) # accuracy_score(Correct answer data,Test results)
print("Correct answer rate:", Accuracy_rate)

[[0, 0], [1, 0], [0, 1], [1, 1]]Prediction result[0 1 1 0]
Correct answer rate: 1.0

I was able to learn using the KNeighborsClassifier algorithm.