search for: hyperplan

...ted from cats data (from R). My dataset corresponds to a completely separable case with a binary response variable ( Sex with 2 levels: F and M), two explanatory variables (Bwt and Hwt) and the classes are balanced. I've used svm() with a linear kernel and I'd like to plot the linear hyperplane and the support vectors. I use plot.svm() and, according to me, I would have found aligned support vectors (because the hyperplane is linear) for each class but it wasn't the case. Could you explain me why ? In addition, when I change the option 'scale' (from TRUE to FALSE) the res...

Dear all, In the svm() function in the package e1071, is there anyway to get the signed distance of a sample point to the separating hyperplane in the feature space? Namely, if the separating hyperplane is given by f(x) = h(x)^T * w - rho, is there any way to get f(x)? Also, in the returned values of the function svm(), what does "$coefs" mean? It is said to be the "corresponding coefficients times the training labels&...

Hello, I am using e1071 to run support vector machine. I would like to plot the data with lattice and specifically show the hyperplanes created by the system. I can store the hyperplane as a contour in an object, and I can plot one object at a time. Since there will be thousands of elements to plot, I can't manually add them one by one to the plot, so I tried to loop into them, but only the last is added. Here it the working...

...ult, crosses indicate support vectors. But why are there two colors of crosses? What do they represent? 2. I want to draw a white-gray colored plot and modify the different colored crosses or circles by different shaped points. Could anyone give me a hint? 3. Is it possible for me to draw a "hyperplane" on the plot? 4. What is the algorithm to plot the contour region? Thank you very much, Frank

On Tue, Oct 27, 2020 at 6:04 PM Luigi Marongiu <marongiu.luigi at gmail.com> wrote: > > Hello, > I am using e1071 to run support vector machine. I would like to plot > the data with lattice and specifically show the hyperplanes created by > the system. > I can store the hyperplane as a contour in an object, and I can plot > one object at a time. Since there will be thousands of elements to > plot, I can't manually add them one by one to the plot, so I tried to > loop into them, but only the last is ad...

...erwise it is easy to achieve via IRLS using ridge regressions. Then even though the data are separated, the penalized log-likelihood has a unique maximum. One intriguing feature is that as the penalty parameter goes to zero, the solution converges to the SVM solution - i.e. the optimal separating hyperplane see http://www-stat.stanford.edu/~hastie/Papers/margmax1.ps -------------------------------------------------------------------- Trevor Hastie hastie@stanford.edu Professor, Department of Statistics, Stanford University Phone: (650) 725-2231 (Statistics)...

...I wonder if there are packages that can do the feature selection and classification at the same time. For instance, I am using SVM to classify my samples, but it's easy to get overfitted if using all of the features. Thus, it is necessary to select "good" features to build an optimum hyperplane (?). Here is a simple example: Suppose I have 100 "useful" features and 100 "useless" features (or noise features), I want the SVM to give me the same results when 1) using only 100 useful features or 2) using all 200 features. Any suggestions or point me to a reference? Than...

...I run this a second time it will be ok. It seems, that kmeans has problems to initialize good starting points, because of the random choose of these starting initial points. With kmeans(data,k,centers=c(...) the problem can be solved. Generally, the starting points can be choose equidistant on a hyperplane of the data, which is also a simple way to get the intitial points (www.fuzzyclustering.de , fc-package of H??ppner, manual). Thank you for your comment, Matthias -----Urspr??ngliche Nachricht----- Von: Unung Istopo Hartanto [mailto:unung at enciety.com] Gesendet: Dienstag, 11. Mai 2004 16:...

...quot;) abline(SimpleLinearReg1, col="red") res <- signif(residuals(SimpleLinearReg1), 2) pre <- predict(SimpleLinearReg1) segments(Age, BloodPressure, Age, pre, col="red") library(calibrate) textxy(Age, BloodPressure, res, cex=0.7) 2) 2) I also need your help plotting a hyperplan for a multiple regression and here I am really stuck... inc <- c(25000, 28000, 17500, 30000, 25000, 32000, 30000, 29000, 26000, 40000) age <- c(60, 36, 21, 35, 33, 43, 29, 45, 41, 48) educ <- c(12, 12, 13, 16, 16, 12, 13, 15, 15, 20) MultLinearReg1 <- lm(inc ~ age + educ) summary(MultLi...

...h datapoint has 19 dimensions. Training with lda(...) works fine, and I'm getting 19 LD coefficients. So far so good. Now I want to visualize the result, and here is where my simple knowledge ends. What I simply want to do is doing a scatterplot on two dimensions and then plotting the projected hyperplane in order to see the dividing line. The scatterplot I'm doing with plot(class1[1:100,1], class1[1:100,2]) points(class2[1:100,1], class2[1:100,2]) but now I face with the problem of determining the intercept for abline(...) (the slope should be -LD[1]/LD[2] I think...). What I really don...

> > So i guess from your prev. email the svmModel$coefs correspond to the > "Alpha" . yes (times the sign of y!). > > Why do I see three columns in the coefs?( Is this the number of classes -1 > = Numbe of hyperplanes) yes, but in a packed format which is not trivial. I attach some explanation I sent to R-help some time ago (the guy wanted to write his own predict method). > Is there a way to determine from the SV's key components of which in this > case would be which of the 2000 genes helped in...

...= trainset, aes(x=x, y=y, color=z)) + geom_point() + scale_color_manual(values = c("red", "blue")) # show support vectors df_sv = trainset[svm_model$index, ] p = p + geom_point(data = df_sv, aes(x=x, y=y), color="purple", size=4, alpha=0.5) # show hyperplane (decision boundaries are off set by 1/w[2]) w = t(svm_model$coefs) %*% svm_model$SV # %*% = matrix multiplication slope_1 = -w[1]/w[2] intercept_1 = svm_model$rho / w[2] p = p + geom_abline(slope = slope_1, intercept = intercept_1, col = "royalblue4") p = p + geom_ribbon(aes(ymin=in...

...gt; > geom_point() + scale_color_manual(values = c("red", "blue")) # show > > support vectors df_sv = trainset[svm_model$index, ] p = p + > > geom_point(data = df_sv, aes(x=x, y=y), > > color="purple", size=4, alpha=0.5) # show hyperplane > > (decision > > boundaries are off set by 1/w[2]) w = t(svm_model$coefs) %*% > > svm_model$SV # %*% = matrix multiplication > > slope_1 = -w[1]/w[2] > > intercept_1 = svm_model$rho / w[2] > > p = p + geom_abline(slope = slope_1, intercept = intercept_1, col =...

..., y=y, color=z)) + > geom_point() + scale_color_manual(values = c("red", "blue")) # show > support vectors df_sv = trainset[svm_model$index, ] p = p + > geom_point(data = df_sv, aes(x=x, y=y), > color="purple", size=4, alpha=0.5) # show hyperplane > (decision > boundaries are off set by 1/w[2]) w = t(svm_model$coefs) %*% > svm_model$SV # %*% = matrix multiplication > slope_1 = -w[1]/w[2] > intercept_1 = svm_model$rho / w[2] > p = p + geom_abline(slope = slope_1, intercept = intercept_1, col = > "royalblue4")...

...e_color_manual(values = c("red", "blue")) # > > > show support vectors df_sv = trainset[svm_model$index, ] p = p + > > > geom_point(data = df_sv, aes(x=x, y=y), > > > color="purple", size=4, alpha=0.5) # show > > > hyperplane (decision boundaries are off set by 1/w[2]) w = > > > t(svm_model$coefs) %*% svm_model$SV # %*% = matrix multiplication > > > slope_1 = -w[1]/w[2] > > > intercept_1 = svm_model$rho / w[2] > > > p = p + geom_abline(slope = slope_1, intercept = intercept_1, col =...

I have a compilation problem on FC2, 2xXeon box. The following dialogue output from the end of the compilation illustrates: [very large snipping sound ...] * DONE (cluster) begin installing recommended package foreign make[2]: *** [foreign.ts] Error 1 make[2]: Leaving directory `/usr/src/redhat/SOURCES/R-2.0.1/src/library/Recommended' make[1]: *** [recommended-packages] Error 2 make[1]:

...8 qhull warning: joggle ('QJ') always produces simplicial output. Triangulated output ('Qt') does nothing. Convex hull of 10 points in 3-d: Number of vertices: 7 Number of facets: 10 Statistics for: | qhull s Qt Tcv QJ Number of points processed: 7 Number of hyperplanes created: 16 Number of distance tests for qhull: 39 CPU seconds to compute hull (after input): 0 Input joggled by: 6.9e-11 Output completed. Verifying that all points are below 6.9e-15 of all facets. Will make 100 distance computations. > -- Daniel E. Bunker TraitNet Associate...

...svm. In general, when an SVM is used for classification, the label of an unknown test-case is decided by the "sign" of its resulting value as calculated from the SVM {-,+} I want the actual values as a proximal representation of the "strength" of the decision. (Further from the hyperplane indicates more confidence) Any suggestions?

...ello, I have noticed a bug with LSSVM implementation in R. It could be a bug with the LSSVM itself that causes this problem. I thought I should post this message to see if anyone else is familiar with this problem and explain why the result is different for odd and even number of cases. Once the hyperplane is found using LSSVM, the prediction results vary when you predict odd or even number of samples. Why? Here I provide e.g. with Iris data in R, keep reducing prediction cases one-by-one, you will see the discrepancy I am talking about. In my own data, this discrepancy between odd and even number o...

...pparently lda from the MASS package can be used in situations with collinear variables. It only produces a warning then but at least it defines a classification rule and produces results. However, I can't find on the help page how exactly it does this. I have a suspicion (it may look at the hyperplane containing the class means, using some kind of default/trivial within-group covariance matrix) but I'd like to know in detail if possible. I find particularly puzzling that it produces different results whether I choose CV=TRUE or I run a manual LOO cross-validation. Constructing an examp...