Question

I have three values that are associated to a specific file. Each file belongs to a specific family. I need to improve a "cross validation" in Weka to understand if with these three values i’m able to identify the family. Now which are the steps to create the Csv or Arff file that i have to open on Weka? My txt has this formatting.

File_Family_1 Value1, Value2,Value3

File_Family_1 Value1, Value2,Value3

File_Family_2 Value1, Value2,Value3

File_Family_2 Value1, Value2,Value3

File_Family_3 Value1, Value2,Value3

File_Family_3 Value1, Value2,Value3

Answer 1

I’ve solved the problem using Excel. In the first row you must write the labels but all in the first column separated by commas.

Family, Attribute1, Attribute2, Attribute3

In the other rows, always all in the first column and separated by commas, you must write the label values.

File_Family1, Value1, Value2, Value3
File_Family2, Value1, Value2, Value3
File_Family3, Value1, Value2, Value3
......

After save as “.Csv” and don’t worry for Excel’s alerts. Open Weka and click on Explorer. So use “Open file” and select the Cvs just created, once opened go in Classify. Finally click on “Choose” to use a specific tree. If start button is disabled it’s need to select a “(Nom)” attribute to perform the “cross-validation”.