At the end of November I blogged about my idea of taking 40 of the highest rated images and turning it into a book. In the post I have described a method for selecting the highest rated images but there were some flaws in my reasoning.
So I took another approach, which is much easier. Let’s assume everyone who has visited my photoblog has voted, but (and this is it) people who didn’t vote actually voted zero stars. In other words, they didn’t found the picture worth at least 1 star. When we assume this, the selection of the images becomes easier because now we only have to select the pictures with the highest total. But to narrow the selection a bit down I have included another rule: an image must have at least 5 votes to participate. Perhaps you’re wondering why I enforce this rule? The reason is quite simple. This way I know at least 5 people found the image good enough to give a rating between 1 – 5 stars.
Furthermore, whenever there is a situation in which two or more images have the same total rating the image with the least votes will be given the highest ranking.
For those of you wondering what the SQL looks like (you do need to have the AjaxRatings addon installed and I assume the prefix of the tables is “pixelpost_“) I have it included below. As can be seen I also let the statement I used earlier (weighted ratings) in the SQL for comparison:
total_votes) as weighedRating
So, how does this look compared to my earlier approach? Well, the top 30 of my images stayed the same, some images shifted a bit but there were no real surprises. At the bottom of the list there were some small changes which has surprised me a bit. For instance: on number 37 (at the time of this post) was the Mosaic image I created containing the 200 images I had on my blog at that time. I’m not sure if I have a high resolution of the image lying around somewhere. In case I don’t I have chosen to omit this file.
Anyway, have a peek at the current output of the script.