Hello, everyone here, is anyone willing to give an instruction on importing the train or test dataset .svm.dat into R? I am newbie in the SVM data format given that I had worked on R for years. I will appreciate any kind help here.
\r\n\r\n
Thanks in advance.
", "messages": [ { "id": "5729", "postDate": "09/29/2011 03:17:54", "content": "Hello, everyone here, is anyone willing to give an instruction on importing the train or test dataset .svm.dat into R? I am newbie in the SVM data format given that I had worked on R for years. I will appreciate any kind help here.
\r\n\r\n
Thanks in advance.
", "rawMarkdown": "", "votes": null }, { "id": "5731", "postDate": "09/29/2011 06:37:30", "content": "http://cran.r-project.org/web/packages/e1071/
\r\n", "rawMarkdown": "", "votes": null }, { "id": "5733", "postDate": "09/29/2011 10:06:07", "content": "
Downloaded, and installed the e1071, kernlab, and klaR packages. Also, downloaded the example1 data from the SVMLight package. Running the SVM on the example1 training data, give an "Error: unexpected symbol . . . ."
\r\nI was expecting that the SVM functions would be able to read the SVMLight data format?
\r\n\r\n\r\nsvmlearn example1/train.dat example1/model
\r\n
\r\nError: unexpected symbol in "svmlearn example1"
\r\nsvmperflearn -c 20 -l 2 --b 0 example1/train.dat example1/model
\r\nError: unexpected numeric constant in "svmperflearn -c 20"
George, I appreciated your help.
\r\nIt seems successful to read the .svmlight.dat with the function "read.matrix.csr" in the package you told "e1071". Eventhougth, the reading finished with warning messages. (What I did: I loaded the "e1071" on R 12.1 and a likily sidekick package "SparseM".\r\n The loading task functioned well given these package are all developed under R 12.2 )
\r\nI did not know the exact looking of the data or R loaded the complete data set which I instructed it to do. Anyway, the loaded train data had 50,000 rows and 999,999 columns. It reasonable for\r\n the data with all labels missing.
", "rawMarkdown": "", "votes": null }, { "id": "5751", "postDate": "09/30/2011 01:49:57", "content": "Hello Beingzy, Can you provide some more detail (maybe the exact command that you used) about reading the data set. I tried the read.matrix.csr command in the R (2.13.1) environment. But, so far, I have been unable to read the file.
\r\nThanks, george
", "rawMarkdown": "", "votes": null }, { "id": "5802", "postDate": "10/01/2011 10:03:28", "content": "> trainFile <- paste(ssfl.data.win, "\\\\public_train_data.svmlight.dat", sep="")
\r\n> trainFile
\r\n[1] "C:\\\\Docume~1\\\\Clyph\\\\MyDocu~1\\\\Work\\\\Analytics\\\\SSFL\\\\data\\\\public_train_data.svmlight.dat"
\r\n> train <- read.matrix.csr(trainFile)
\r\nWarning message: closing unused connection 3
\r\n(C:\\Docume~1\\Clyph\\MyDocu~1\\Work\\Analytics\\SSFL\\data\\public_train_data.svmlight.dat)
\r\n> ls()
\r\n[1] "bmi" "height" "hh" "oops" "rrr" "ssfl.data.win" "train"
\r\n[8] "train.file" "trainFile" "vowel.train" "weight" "x" "xbar"
http://cran.r-project.org/web/packages/e1071/
\r\n", "votes": null, "replies": [] }, { "id": 5733, "author_name": "george11", "author_url": "", "post_date": "09/29/2011 10:06:07", "content": "
Downloaded, and installed the e1071, kernlab, and klaR packages. Also, downloaded the example1 data from the SVMLight package. Running the SVM on the example1 training data, give an "Error: unexpected symbol . . . ."
\r\nI was expecting that the SVM functions would be able to read the SVMLight data format?
\r\n\r\n\r\nsvmlearn example1/train.dat example1/model
\r\n
\r\nError: unexpected symbol in "svmlearn example1"
\r\nsvmperflearn -c 20 -l 2 --b 0 example1/train.dat example1/model
\r\nError: unexpected numeric constant in "svmperflearn -c 20"
George, I appreciated your help.
\r\nIt seems successful to read the .svmlight.dat with the function "read.matrix.csr" in the package you told "e1071". Eventhougth, the reading finished with warning messages. (What I did: I loaded the "e1071" on R 12.1 and a likily sidekick package "SparseM".\r\n The loading task functioned well given these package are all developed under R 12.2 )
\r\nI did not know the exact looking of the data or R loaded the complete data set which I instructed it to do. Anyway, the loaded train data had 50,000 rows and 999,999 columns. It reasonable for\r\n the data with all labels missing.
", "votes": null, "replies": [] }, { "id": 5751, "author_name": "george11", "author_url": "", "post_date": "09/30/2011 01:49:57", "content": "Hello Beingzy, Can you provide some more detail (maybe the exact command that you used) about reading the data set. I tried the read.matrix.csr command in the R (2.13.1) environment. But, so far, I have been unable to read the file.
\r\nThanks, george
", "votes": null, "replies": [] }, { "id": 5802, "author_name": "george11", "author_url": "", "post_date": "10/01/2011 10:03:28", "content": "> trainFile <- paste(ssfl.data.win, "\\\\public_train_data.svmlight.dat", sep="")
\r\n> trainFile
\r\n[1] "C:\\\\Docume~1\\\\Clyph\\\\MyDocu~1\\\\Work\\\\Analytics\\\\SSFL\\\\data\\\\public_train_data.svmlight.dat"
\r\n> train <- read.matrix.csr(trainFile)
\r\nWarning message: closing unused connection 3
\r\n(C:\\Docume~1\\Clyph\\MyDocu~1\\Work\\Analytics\\SSFL\\data\\public_train_data.svmlight.dat)
\r\n> ls()
\r\n[1] "bmi" "height" "hh" "oops" "rrr" "ssfl.data.win" "train"
\r\n[8] "train.file" "trainFile" "vowel.train" "weight" "x" "xbar"