Machine Learning Artificial Intelligence ML Statistics 2 Supervised

기계학습의 종류 • Supervised Learning • Unsupervised Learning • Reinforcement Learning 3

회귀 모형 • linear regression / partial least square • penalized regression(LASSO, ridge, elastic

분류 모형 - Linear Discriminant Analysis / Quadratic Discriminant Analysis - Logistic regression -

평가 지표 예측 실제 Positive Negative Positive True Postive False Negative (Type II error)

Precision Positive Predictive Value 예측 실제 Positive Negative Positive True Postive False Negative (Type

Recall True Positive Rate, Sensitivity 예측 실제 Positive Negative Positive True Postive False Negative

Specificity True Negative Rate 예측 실제 Positive Negative Positive True Postive False Negative (Type

Accuracy 예측 실제 Positive Negative Positive True Postive False Negative (Type II error) Negative

Kappa O : observed accuracy, E : Expected Accuracy 35

오늘 다룰 모형 ##이산/범주 변수를 다루자 - Linear Discriminant Analysis - Logistic regression(Elastic net)

CARET R package CARET(Classicifation And REgression Training) - 매우 다양한 모형을 하나의 방식으로 분석하게

Elastic Net - Lasso + Ridge - Tuning parameter : λ, α λ :

Slides: 53

Download presentation

Machine Learning Artificial Intelligence ML Statistics 2

기계학습의 종류 • Supervised Learning • Unsupervised Learning • Reinforcement Learning 3

회귀 모형 • linear regression / partial least square • penalized regression(LASSO, ridge, elastic net) • Neural network • Multivariate Adaptive regression Splines • Support Vector Machine • K-Nearest Neighbors • Regression tree / Random forest • . . .

분류 모형 - Linear Discriminant Analysis / Quadratic Discriminant Analysis - Logistic regression - Nearest shrunken centroids - Neural Network - Flexible discriminant analysis - Support Vector Machine - Naive Bayes - Classification tree / Random forest -. . .

평가 지표 예측 실제 Positive Negative Positive True Postive False Negative (Type II error) Negative False Positive (Type I error) True Negative 30

Precision Positive Predictive Value 예측 실제 Positive Negative Positive True Postive False Negative (Type II error) Negative False Positive (Type I error) True Negative 31

Recall True Positive Rate, Sensitivity 예측 실제 Positive Negative Positive True Postive False Negative (Type II error) Negative False Positive (Type I error) True Negative 32

Specificity True Negative Rate 예측 실제 Positive Negative Positive True Postive False Negative (Type II error) Negative False Positive (Type I error) True Negative 33

Accuracy 예측 실제 Positive Negative Positive True Postive False Negative (Type II error) Negative False Positive (Type I error) True Negative 34

Kappa O : observed accuracy, E : Expected Accuracy 35

F 1 score 36

Area Under Curve 37

모형 선택 Cross-Validation 38

오늘 다룰 모형 ##이산/범주 변수를 다루자 - Linear Discriminant Analysis - Logistic regression(Elastic net) - K-Nearest Neighbors - Neural Network - Support Vector Machine - Naive Bayes - Classification tree / Random forest

CARET R package CARET(Classicifation And REgression Training) - 매우 다양한 모형을 하나의 방식으로 분석하게 도와주는 패키 지

Elastic Net

Lasso vs. Ridge

Elastic Net - Lasso + Ridge - Tuning parameter : λ, α λ : bias의 양 α : 두 가지 bias의 종류의 비중

Linear Discriminant Analysis

Support Vector Machine Kernel Parameter