<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">    Good question, Jack!<br>
      <br>
          You're missing some information here.  As you know,
      statistical significance tells you whether your sample is large
      enough that the effect you observe is probably not due to a bad
      sample.<br>
      <br>
      <a class="moz-txt-link-freetext" href="http://grieve-smith.com/blog/2014/01/how-big-a-sample-do-you-need/">http://grieve-smith.com/blog/2014/01/how-big-a-sample-do-you-need/</a><br>
      <br>
          So you've figured out what you want to sample (for example, a
      language, genre or subgenre) and taken a representative sample of
      it.  You must therefore know the size of the sample.  That is your
      <i>n</i>, which you will use to calculate the statistical
      significance of the difference between the two values, probably
      with Student's <i>t</i>-test.<br>
      <br>
          It is the <i>t-</i><tt>test that will tell you whether </tt>your
      sample is too small to rule out the possibility that you just
      sampled the wrong things.  You also need your hypothesis: if you
      were expecting the values of System 02 to be higher, you want a
      one-tailed <i>t</i><tt>-test; otherwise you want a two-tailed
        test.</tt><br>
      <br>
          If you don't actually have a representative sample or a
      hypothesis, then you're just playing an elaborate game of pretend
      with your reviewers, where you pretend to find significance and
      they pretend to be impressed.  In that case, I'm not sure exactly
      what you have to pretend to do; maybe someone else can fill us in.<br>
      <br>
<a class="moz-txt-link-freetext" href="http://grieve-smith.com/blog/2014/01/you-cant-get-significance-without-a-representative-sample/">http://grieve-smith.com/blog/2014/01/you-cant-get-significance-without-a-representative-sample/</a><br>
      <br>
      On 11/10/2014 6:28 AM, Jack Alan wrote:<br>
    </div>
    <blockquote
cite="mid:CAPgQO-L8Vi6NDrbAUfeLdEavtrbw3jJ+UWjj-oz2uZonLupyEQ@mail.gmail.com"
      type="cite">
      <div dir="ltr">Hi folks,
        <div><br>
        </div>
        <div>A bit struggling of calculating the statistical significant
          between the output of two systems. Suppose Ive got the
          following two results from two independent systems (performing
          sequence labelling task):</div>
        <div><br>
        </div>
        <div>System 01: </div>
        <div>precision:  81.57%; recall:  57.12%; FB1:  67.19%<br>
        </div>
        <div><br>
        </div>
        <div>
          <div>System 02: </div>
          <div>precision:  84.07%; recall:  62.47%; FB1:  71.68%</div>
        </div>
        <div><br>
        </div>
        <div><br>
        </div>
        <div>Could someone pinpoint me to the way of calculating the
          statistical significant between them?</div>
        <div><br>
        </div>
        <div>p.s. I've no folds applied (just one go "training and
          test")</div>
        <div><br>
        </div>
        <div>J.</div>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
UNSUBSCRIBE from this page: <a class="moz-txt-link-freetext" href="http://mailman.uib.no/options/corpora">http://mailman.uib.no/options/corpora</a>
Corpora mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Corpora@uib.no">Corpora@uib.no</a>
<a class="moz-txt-link-freetext" href="http://mailman.uib.no/listinfo/corpora">http://mailman.uib.no/listinfo/corpora</a>
</pre>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
                                -Angus B. Grieve-Smith
                                <a class="moz-txt-link-abbreviated" href="mailto:grvsmth@panix.com">grvsmth@panix.com</a>
</pre>
  </body>
</html>