Generally, you can do your analysis and plotting using WEKA. In this case,
in the dataset file, you need to have the whole features (attributes) as
columns for each instance/case (row) and *one* associated class label to be
predicted. WEKA can calculate and plot some evaluation metrics, e.g, ROC
On the other hand, I don't think that you can do much with the stuff you
provided. Thus, to work with WEKA, you need to create your dataset again
based what I mentioned previously. In addition, it seems like you're
working on multi-label problem (precisely multi-target problem) so you need
to be careful with this issue and keep only one class to be predicted.
Otherwise, use MEKA tool.
On 22 Sep 2016 1:48 pm, "Reena [via WEKA]" <
I have the following data in excel. I have manually extracted clinical
procedures from patient discharge summary which is in free-form
text(actual) as well as using an automated tool (predicted). I need to find
out how accurate the automated tool is. First 2 columns are actual. Next 2
columns are predicted. I have also calculated tp, fp and fn. My questions
1. What type of analysis can be done for this type of data? In WEKA or
Excel? I can not draw ROC or PRC as I don't have the probabilities.
2. What kind of charts can be plotted for this? I have calculated
Precision, Recall and F-score, but these are just 3 numbers. What type of
chart can I draw?
I will really appreciate your help on this.
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/
List etiquette: http://www.cs.waikato.ac.nz/~
If you reply to this email, your message will be added to the discussion
To unsubscribe from WEKA, click here