Exam review

[jhooks1]

Regarding KNN on the review (Sept 10):

"Make the case the Knn suffers from the following problems:

• They are computationally expensive (in terms of run time and memory)
• They are intolerant of attribute noise.
• They are intolerant of irrelevant attributes.
• They are sensitive to the choice of the algorithm's similarity function.
• They provide little usable information regarding the structure of the data
• There is no natural way to work with nominal-valued attributes or missing attribute"

If we get this question on the exam, are we expected to just list these problems, or do we need to prove them somehow?

[Tim Menzies]

thanks for asking this question on the forum. that way, we can all see the question and help with the answer.

I'd add one "for example" to each line. e.g. "They provide little usable information regarding the structure of the data". For example, while other learners offer some statement about patterns in the data (such as the frequency counts of Naive Bayes), all a standard kNN does is store the training examples in a box with no summary information.

--
You received this message because you are subscribed to the Google Groups "csx73" group.
To unsubscribe from this group and stop receiving emails from it, send an email to csx73+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--

:: there are some who call me...   tim.m...@gmail.com

:: prof @ cs.ai.se.csee.wvu.usa.sol.virgo.all.nil

:: +1-304-376-2859          

:: http://menzies.us      (skype = menzies.tim)

<hubris>

  vita=   http:// goo.gl/8eNhY  

  pubs=   http:// goo.gl/8KPKA

  stats=   http:// goo.gl/vggy1 

  wow =   http:// goo.gl/2Wg9A

</hubris>

[jhooks1]

Thanks. 

I have some more questions:

"computer the variance of 6 tens and 2 gives."  Are gives zeros (and a zero variance)?


From the comparing different learners lecture:

one  :mu 0.3 :rank 1
two  :mu 0.8 :rank 2

We are displaying these results in a percentile chart from .45 to .85. 
How is .3 not getting cut off?

[Tim Menzies]

On Sat, Oct 5, 2013 at 11:34 PM, jhooks1 <jho...@mix.wvu.edu> wrote:
> Thanks.
>
> I have some more questions:
>
> "computer the variance of 6 tens and 2 gives." Are gives zeros (and a zero
> variance)?

6 6 6 6 6 6 6 2

mu = (6*6 +2)/6 = 38/7 = 5.42

sum (x_i - mu)^2 = 6*(6-6.33333)^2 + (2-6.333)^2 = 6*0.333^2 +
-4.333^2 = 0.666+ 16 = 13.73

sd = sqrt(sum/(n-1)) = sqrt(13.73/(7-1)) = 1.51

> From the comparing different learners lecture:
>
> one :mu 0.3 :rank 1
> two :mu 0.8 :rank 2
>
> We are displaying these results in a percentile chart from .45 to .85.
> How is .3 not getting cut off?

my bad. typo. the range should be wider

t

[jhooks1]

Are we allowed to use calculators on this exam (e.g. calculating logarithms)?

[Tim Menzies]

you can use calculators or leave your answers as fractions.

t

On Mon, Oct 7, 2013 at 3:49 PM, jhooks1 <jho...@mix.wvu.edu> wrote:

Are we allowed to use calculators on this exam (e.g. calculating logarithms)?

--

You received this message because you are subscribed to the Google Groups "csx73" group.
To unsubscribe from this group and stop receiving emails from it, send an email to csx73+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Jason Hooks]

Thanks for answering my questions.

Can you go over

  happy? age       hair
  ------ --------  ------
  y      teenager  black
  y      teenager  bald
  y      teenager  blond
  y      teenager  blond
  y      old man   black
  n      old man   blond
  n      old man   bald

• In the following example, what is the entropy of class variable happy? in the entire data set? (see the algorithm later in the section

I am a little confused on the algorithm and what the answer is to the entire data set.  Entropy is listed as E(r) = sum Pi*log(Pi).  should it be sum -Pi*log(Pi)?

[Jeffrey Yancey]

Thanks for all the examples, they've been helpful while studying.

I would just like to echo JHook's question concerning the following: "computer the variance of 6 tens and 2 gives.". Specifically asking what a "give" is. I have no questions regarding the computation, that is quite clear.

Again, I really appreciate this group's discussion.

[Brian Dunar]

I think he means calculate the entropy of...

10,10,10,10,10,10, 2

-Brian

[Brian Dunar]

Sorry, I meant variance.

[Tazin Afrin]

I think it makes more sense if it is "computer the variance of 6 tens and 2 fives"

then it would be the variance of 10 10 10 10 10 10 5 5 

:)

[Jeffrey Yancey]

Tazin, you are a genius.