Educational Application For Neural Network - Correlate Entry Test and Single Skill Level

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Educational Application For Neural Network - Correlate Entry Test and Single Skill Level

kickokim
Hello, I'm a English as a foreign language teacher and don't know anything about programming and I only know about neural network from articles I've found on the web!
I'm curious to know if it's possible to set up a neural network to correlate a whole test made up of multiple choice quizzes  and a single result that will be the skill level of the student (A1/A2/B1/B2 level and so on). This could help teachers and schools sort students together in the right classes.
At the moment these types of tests are not very efficient cause the results are given only by the percentage of right answers to the multiple choice quizzes, but teachers also know that the type of mistake a student makes also varies from level to level.  
Could a neural network help give better level assessment?

Thank you in advanced

Kim
Reply | Threaded
Open this post in threaded view
|

Re: Educational Application For Neural Network - Correlate Entry Test and Single Skill Level

jyoshimi
Administrator
Hi there,

One thing you could try is an unsupervised method, like the self organizing map or the competitive network.   You'd have to convert each test into a binary valued vector, e.g.

1:A, 2:B, 3:D,...

would become

1,0,0,0 ,0,1,0,0 ,0,0,0,1, ...

Then feed those to an unsupervised network and see how it clusters things.  If there are patterns in the data independently of the score on the test, e.g. different patterns of errors, it might show up.

 If you are comfortable with python you could also try one of these methods

http://scikit-learn.org/stable/unsupervised_learning.html

I should probably make a video about SOM, competitive, etc...

- Jeff