members_score not always weighted

Issue #100 resolved
Dmitry created an issue

Doc states:

members_score: A weighted score based on the rating MyAnimeList members have given the title. Only available in anime detail requests.

For example for http://myanimelist.net/anime/31017 API returns 10 and it is not weighted score. I think API should return 0 in this case cause most of the time not weighted scores on MAL are useless.

Comments (6)

  1. Michael Johnson

    You're free to ignore the data point if you don't find it useful. The score for that title across members is 10, just 1 member rated it and so that's the weighting in that case.

    Feel free to suggest to MAL they do something about the data they provide.

  2. Dmitry reporter

    Well, they do. For example http://myanimelist.net/anime/31017 (notice that small text says it's not weighted)

    <span>10.00</span><sup>1</sup> (scored by <span>1</span> users)
          <div class="statistics-info info1">
        <small><sup>1</sup>
            This score is <span style="text-decoration: underline;">not</span> weighted</span>. Please note that 'Not yet aired' titles are excluded.
        </small>
      </div>
    

    And http://myanimelist.net/anime/21 (notice that it has itemprop="ratingValue" and small text says it's weighted)

    <span itemprop="ratingValue">8.60</span><sup>1</sup> (scored by <span itemprop="ratingCount">268,997</span> users)
          <meta itemprop="bestRating" content="10">
        <meta itemprop="worstRating" content="1">
          <div class="statistics-info info1">
        <small><sup>1</sup>
            indicates a weighted score. Please note that 'Not yet aired' titles are excluded.
        </small>
    

    So, to get only weighed score we can just take it from itemprop="ratingValue".

  3. Ratan Dhawtal

    I think he has a point.

    In this case the docs or endpoint must be adjusted. Ratings based on 1 user while the serie did not aired yet are indeed useless.

    @motokochan I think that we should adjust the object parsing way which @Dmitry recommended.

  4. Michael Johnson

    The code on the site changed, but I went ahead and am now ignoring any scores that are not weighted.

  5. Log in to comment