MetaPost comments and questions about the messageboards and other parts of EN World. If you have a problem, this is where to go. Moderator email addresses are posted here.
I thought I would cull this string of posts from the early thread as it has gotten lost in the shuffle, I think.
Quote:
Originally Posted by Mark
Not sure if this has been addressed but it appears that products that haven't been reviewed are treated as having no stars and thus bring down the rating of a publisher.
Quote:
Originally Posted by Michael Morris
Rats.. Hmm.. Well... I uh.. As soon as I figure out a correction to the query I'll recalculate it.
Quote:
Originally Posted by jmucchiello
Shouldn't you just average the existing reviews without regard for how many products there are?
Select Sum(stars) as total_starts, Count(*) as total_reviews, publisher_id from reviews; (or join the reviews and pubs table if the pub_id is not available on the review table)
SELECT AVG(score) AS score FROM product WHERE publisherid = $publisherid AND score > 0
Haven't had time to test it yet - been taking a break.
Quote:
Originally Posted by jmucchiello
But that averages the product scores which are already an average. So if a publisher have two products: one with 100 reviews and an average of 3 and one with 1 review of 5, the publisher will have a score of 4 by that querry. That's why my querry was off of the (assumed) review table instead of the product table.
Thanks Mark. I think other items like having people be able to add products and getting the reviews sortible by reviewer and getting the Mods our ability to more easily function over there have taken a slight priority.
Took a break from working with the reviews system to make some changes Russ requested for the downloads section.
OK. Here's the problem and why I think you need to prioritize the "Publisher's Ratings in the Reviews Section" problem higher, at least insofar as addressing it in some way even if you cannot fix it. It is currently giving out FALSE information and potentially damaging the reputations of publishers who work very hard to produce quality products. I understand you are quite busy and perhaps have a different idea of what is important. That's fine. I don't expect you to hold the same opinions as I do. However, if you can't fix them, the least you can do is to take them offline until they are correct.
Ok it's fixed. For those who know PHP and SQL - here's the algorythm.
Code:
$getproducts = $DB_site->query("
SELECT productid as id from product
");
while ($product = $DB_site->fetch_array($getproducts))
{
$getproductavg = $DB_site->query_first("
SELECT AVG(score) AS score
FROM review
LEFT JOIN post as post ON(post.postid = review.reviewid)
WHERE post.threadid = $product[id]
AND score > 0
");
$productscore = round($getproductavg['score']);
// Insert it.
$DB_site->query("
UPDATE product
SET weightscore = $productscore
WHERE productid = $product[id]
");
// Find Publisher ID
$getpublisherid = $DB_site->query_first("
SELECT publisherid
FROM product
WHERE productid = $product[id]
");
$publisherid = $getpublisherid['publisherid'];
// Find the Publisher's average product rating.
$getpublisheravg = $DB_site->query_first("
SELECT AVG(weightscore) AS score
FROM product
WHERE publisherid = $publisherid
");
$publisherscore = round($getpublisheravg['score']);
// Insert it.
$DB_site->query("
UPDATE publisher
SET avgscore = $publisherscore
WHERE publisherid = $publisherid
");
}
No, it isn't. Somehow, and I won't pretend to follow the code since it is not my area, you have the total of review scores divided total products, even when some products aren't reviewed. It's my understanding that a Publisher's rating should simply be total reviews averaged, or perhaps averaged for each product, and then averaged again and divided by the number of products reviewed (though this latter method can be decieving if one product has only one review and it is reviewed to an extreme, while other products received multiple reviews that are then more indicative of an actual average).
If one has a three star review, a four star review and a five star review, they have a Publisher rating of four stars, regardless of how many products they have in total and regardless if some haven't been reviewed at all. Currently, it would seem, if a publisher had the above suggested reviews, perhaps all for a single product, but also had eleven additional products (for a total of twelve products) they'd have a publisher rating of one star (three stars plus four stars plus five stars equal twelve stars divided by twelve products equal a rating of one star)
You could look at Dragon Scale Counters too. We have all good reviews (4-5 stars) yet when you look at the overall, we have a 2. Furthermore, until Crothian did his last review I think it showed us with NO stars, even though we had 3 other reviews that were good.
While Im here, any ETA on when we can get our entry for the company edited? It used to have a picture, point correctly to our page, and have a blurb. Now the blurb is gone, the image is gone, and the link to the homepage has an extra http:// in it.
Since I know that CMG has had no reviews below a 3 star rating, and actually is hovering around four stars, because of a solid number of four star reviews and a good number of five star reviews, I've been perplexed that my company has a two star rating by the new system.
Maybe you should create a couple of "control" companies with an easy-to-figure-out Publisher rating (like the one in my example above) so that your system won't fool you into believing it is fixed when it is not.
If you're going to start taking requests, please start a new thread so that this subject isn't muddied yet again. I've already had to relocate the matter from one other over-worked thread (and the BUG has been pronounced fixed once, when it wasn't), so I'd rather not have the issue confused even more, please.
Anyway, I will have a 2 week break to work on this - I intend to have everything trounced within that time.
Quote:
Originally Posted by Michael Morris (02-26-05)
Mark, I've begun work on the admin section of the reviews site. I should have this information fixed shortly.
Quote:
Originally Posted by Michael Morris (03-04-05)
It will be fixed today - specifically tonight.
Quote:
Originally Posted by Michael Morris (03-06-05)
Were's this happening Mark? It'd be useful to watch that specific publisher while I tinker with it to make sure it corrects out.
OK. In addition to the other posts I've made regarding my observations and suggestions, I want to point out that even if you were to "watch that specific publisher" (CMG) while you "tinker with it to make sure it corrects out", you will notice no change from my perspective so I am not the one to ask. All I have access to are the typical pages that anyone visiting the site can see and there has been no change in what I have seen. This has been frustrating, to say the least. I'm not asking for any more promises or even estimates of when this will be corrected, but I would appreciate some answers to the following questions...
1.) Can you remove that aspect of the reviews section (Publisher Rating) until you actually have time to work on it?
2.) Can you give publishers the ability to edit their own sections, so that there is less work for you in regard to non-code areas?
3.) Can you find some way to delegate some of the bugs/fixes that need handling off to other people so that you aren't overtaxed?
Linear Mode
And yet another word from our sponsors
