cjlin1
commented
5 years ago It does have negative values:
$ cut -f 4 -d ' ' train3rd.data | cut -c 3-|sort -n -4293791624 -4293789819 -4293777007 ... 2028513 2041403 2042181
Xinyao90 writes:
I use svm-scale to scale my input so that each feature has range in [0,1] and the parameters are stored in file rang3rd. However, the minimum value of the third feature is wrong since the values of the third feature are all positive but the minimum value is negative.
The command I used: svm-scale -l 0 -u 1 -s range3rd train3rd.data > train3rd.scale Here is the content in range3rd file, the 3rd feature's min negative value is in bold: x 0 1 1 626564 792386 2 216271 277056 3 -4293791624 2042181 4 64541 82353 5 5585 13492 6 524523 1020279 7 0 11
TrainingData&RangeFile.zip
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.*
I use svm-scale to scale my input so that each feature has range in [0,1] and the parameters are stored in file rang3rd. However, the minimum value of the third feature is wrong since the values of the third feature are all positive but the minimum value is negative.
The command I used: svm-scale -l 0 -u 1 -s range3rd train3rd.data > train3rd.scale Here is the content in range3rd file, the 3rd feature's min negative value is in bold: x 0 1 1 626564 792386 2 216271 277056 3 -4293791624 2042181 4 64541 82353 5 5585 13492 6 524523 1020279 7 0 11
TrainingData&RangeFile.zip