Reformat csv reading and plot the number of instances being used throughout the hypothesis validation process

[dpuenteramirez]

The basic idea is to add a new metric on the right of the current plot. A handmade of the desired result is the following.

[dpuenteramirez]

The desired result has been achieved

[alvarag]

I am not sure about the names... I think we should have:

• Original dataset
• After self-training
• After ENN

[dpuenteramirez]

You mean on the right legend, is that correct?

[dpuenteramirez]

Something like this @alvarag?

[alvarag]

Yes, that is what I wanted to see. The problem is that ENN is fitlering instances of the original dataset, thus it is worsening the performance

[dpuenteramirez]

Yeah, I knwon, in the following picture you can see the same experiment without deletion of the originl dataset instances. Thus the performance improves quite a bit

[dpuenteramirez]

Nevertheless, both the "Banana" and the "Iris" datasets have been used as tests to check that all the code worked correctly. I am currently running the experimentation on all the datasets we have in order to have a wider range of results.