motion-picture demo credit: https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjsf1EEvTU1tDDMHW7swGBpfieFBhAEqRIMP7LyICWZ3Q3VCKzIJc-ksGkTsfmch5vaUkt16ciBR4794iDrQmw780IrYXZMNE0vjdjNX-d_VGvm9EbEieYV6UcyS-aBohRxGkauMDqQNpME/s1600/SVM+with+r.png
____________________________________________________________________
Code Credit:
https://github.com/deltaphase/SCUPSY_statistical_computing_R_2017/blob/master/R/machine%2Blearning%2BSVM%2BSVR.r
| library("e1071") | |
| head(iris,5) | |
| pairs(iris[,1:4], col=as.integer(iris[,5])+1) | |
| #str(iris) | |
| #unique(iris$Species) | |
| #summary(iris) | |
| attach(iris) | |
| detach(iris) | |
| x <- subset(iris, select=-Species) | |
| y <- Species | |
| #svm_model <- svm(Species ., data=iris) | |
| svm_mode1 <- svm(x,y, cost = 1, gamma = 0.25) | |
| summary(svm_mode1) | |
| pred <- predict(svm_mode1,x) | |
| table(pred,y) | |
| svm_mode1 <- svm(x,y, cost = 10, gamma = 2) | |
| pred <- predict(svm_mode1,x) | |
| table(pred,y) | |
| # Visualizing SVM model (1) | |
| svm_mode1 <- svm(Species ., data=iris, cost = 1, gamma = 0.25) | |
| plot(svm_mode1, iris, Petal.Width Petal.Length, | |
| slice = list(Sepal.Width = 3, Sepal.Length = 4),color.palette = terrain.colors) | |
| # Visualizing SVM model (2) | |
| plot(svm_mode1, iris, Sepal.Width Petal.Width, | |
| slice = list(Sepal.Length = 3, Petal.Length = 4),color.palette = terrain.colors) | |
| svm_tune <- tune(svm, train.x=x, train.y=y, | |
| kernel="radial", ranges=list(cost=10^(-1:2), gamma=c(.5,1,2))) | |
| print(svm_tune) | |
| svm_model_after_tune <- svm(Species ., data=iris, kernel="radial", | |
| cost=1, gamma=0.5) | |
| summary(svm_model_after_tune) | |
| pred <- predict(svm_model_after_tune,x) | |
| table(pred,y) | |
| # SVR | |
| # https://github.com/alexandrekow/svmtutorial/tree/master/SupportVectorRegressionInR | |
| #### Step 1, regression | |
| # Load the information from the csv file | |
| #dataDirectory <- "D:/DropBox/Dropbox/@Alex_Tutoriaux/" | |
| dataDirectory <- "/Users/kevinhsu/Downloads/" | |
| data <- read.csv(paste(dataDirectory, 'regression.csv', sep=""), header = TRUE) | |
| # Plot the information | |
| plot(data, pch=16) | |
| # Create a linear regression model | |
| model <- lm(Y X, data) | |
| # Add the fitted line | |
| abline(model) | |
| #### Step 2, regression, RMSE | |
| plot(data, pch=16) | |
| model <- lm(Y X , data) | |
| # brand a prediction for each X | |
| predictedY <- predict(model, data) | |
| # display the predictions | |
| points(data$X, predictedY, col = "blue", pch=4) | |
| # This role volition compute the RMSE | |
| rmse <- function(error) | |
| { | |
| sqrt(mean(error^2)) | |
| } | |
| error <- model$residuals # same equally data$Y - predictedY | |
| predictionRMSE <- rmse(error) # 5.703778 | |
| #### Step 3, regression vs SVR | |
| rmse <- function(error) | |
| { | |
| sqrt(mean(error^2)) | |
| } | |
| plot(data, pch=16) | |
| # linear model ============================================== | |
| model <- lm(Y X , data) | |
| predictedY <- predict(model, data) | |
| points(data$X, predictedY, col = "blue", pch=4) | |
| error <- model$residuals # same equally data$Y - predictedY | |
| predictionRMSE <- rmse(error) # 5.703778 | |
| # destination of linear model ======================================= | |
| plot(data, pch=16) | |
| # svr model ============================================== | |
| if(require(e1071)){ | |
| model <- svm(Y X , data) | |
| predictedY <- predict(model, data) | |
| points(data$X, predictedY, col = "red", pch=4) | |
| # /!\ this fourth dimension svrModel$residuals is non the same equally data$Y - predictedY | |
| # then nosotros compute the mistake similar this | |
| error <- data$Y - predictedY | |
| svrPredictionRMSE <- rmse(error) # 3.157061 | |
| } | |
| # destination of svr model ======================================= | |
| #### Step 4, tuning SVR | |
| rmse <- function(error) | |
| { | |
| sqrt(mean(error^2)) | |
| } | |
| plot(data) | |
| # linear model ============================================== | |
| model <- lm(Y X , data) | |
| predictedY <- predict(model, data) | |
| points(data$X, predictedY, col = "blue", pch=4) | |
| error <- model$residuals # same equally data$Y - predictedY | |
| predictionRMSE <- rmse(error) # 5.703778 | |
| # destination of linear model ======================================= | |
| plot(data) | |
| # svr model ============================================== | |
| if(require(e1071)){ | |
| model <- svm(Y X , data) | |
| predictedY <- predict(model, data) | |
| points(data$X, predictedY, col = "red", pch=17) | |
| error <- data$Y - predictedY # /!\ this fourth dimension svrModel$residuals is non the same equally data$Y - predictedY | |
| svrPredictionRMSE <- rmse(error) # 3.157061 | |
| tuneResult <- tune(svm, Y X, data = data, | |
| ranges = list(epsilon = seq(0,1,0.1), cost = 2^(2:9)) | |
| ) | |
| print(tuneResult) # best performance: MSE = 8.371412, RMSE = 2.89 epsilon 1e-04 toll 4 | |
| # Draw the commencement tuning graph | |
| plot(tuneResult) | |
| # On the commencement tuning graph, nosotros tin encounter that the graph is darker on the leftside when epsilon is small, | |
| # then nosotros suit the tuning to acquire inward this direction | |
| # Draw the instant tuning graph | |
| tuneResult <- tune(svm, Y X, data = data, | |
| ranges = list(epsilon = seq(0,0.2,0.01), | |
| cost = 2^(2:9)) | |
| ) | |
| print(tuneResult) | |
| plot(tuneResult) | |
| plot(data, pch=16) | |
| tunedModel <- tuneResult$best.model | |
| tunedModelY <- predict(tunedModel, data) | |
| points(data$X, predictedY, col = "red", pch=4) | |
| lines(data$X, predictedY, col = "red", pch=4) | |
| points(data$X, tunedModelY, col = "blue", pch=4) | |
| lines(data$X, tunedModelY, col = "blue", pch=4) | |
| error <- data$Y - tunedModelY | |
| # this value tin move dissimilar because the best model is determined yesteryear cross-validation over randomly shuffled information | |
| tunedModelRMSE <- rmse(error) # 2.219642 | |
| } | |
| # destination of svr model ======================================= | |
| tuneResult$best.model |