I write a memo about the classification of supervised learning, which I mainly studied using Aidemy.
--Find and classify the boundaries of linearly separable data by learning
--The border becomes a straight line --Used for data with few classes such as binary classification --Computable probability of data being classified into classes (precipitation probability, etc.)
--Cannot be classified unless it is linearly separable --Not suitable for high-dimensional sparse data --Borderline is not versatile (pulled by training data)
from sklearn.linear_model import LogisticRegression
--scikit-learn library
--linear_module submodule
--LogisticRegression () function
--Classified by data boundaries --SVM that draws a straight line to classify
--Support vector (data group close to other classes) --Draw a border at the position where the distance from the support vector is the largest --Easy to generalize and improve data classification prediction --Easy to make a line
――Forecast tends to be slow as the amount of data increases --Cannot be classified unless it is linearly separable
from sklearn.svm import LinearSVC
Recommended Posts