What should be default value of features for classification stored in csv file using svm classifciation

asked 2017-05-01 03:04:56 -0600

vidushig2 gravatar image

updated 2017-05-01 07:13:17 -0600

Hi, I am writing code in cpp. My task is to classify part of document image belongs to table or not.I have dataset consisting of 80 items.20 items have 20 features each.The value of each feature is stored in csv file.60 items have 26 features and their value is also stored in same csv file.Now 20 items have 20 features corresponding to them,and 60 items have 26 features.What should be the default value of 6 features of 20 items who have 20 .

edit retag flag offensive close merge delete

Comments

can you maybe explain, how that happened ?

what are you trying to achieve, in general ? (80 samples is "next to nothing" for training an SVM)

berak gravatar imageberak ( 2017-05-01 07:28:58 -0600 )edit

Let me explain it again.There are 100 items that need to be classified using svm classification.Each items have some feature value corresponding to them.Now 75 items have 26 features and their corresponding value.Remaining 25 items have only 20 features and their corresponding value.These 100 items with their features are saved in csv file.What should be the default value of remaining 6 features of 25 items.There are 26 columns in csv file and 100 rows.75 items have all the 26 columns filled with values.25 items have 20 columns filled with values.What should be the default value of 6 columns of 25 items which have only 20 features.Should I write default value as 0.000000 or -1.000000 for features that are not part of 25 items.

vidushig2 gravatar imagevidushig2 ( 2017-05-01 07:43:24 -0600 )edit

Someone please help in this question

vidushig2 gravatar imagevidushig2 ( 2017-05-02 00:42:57 -0600 )edit