SELECT pageid, rank, url, bytesJS, visualComplete FROM [httparchive:runs.latest_pages] WHERE rank < 10000 AND visualComplete != 0 AND bytesJS != 0 ORDER BY visualComplete desc
Hmm… I don’t think the above query actually answer the question. We would need at least another level of aggregation to bucket the amount of downloaded JS and then see if there is a correlation between these buckets and visualComplete.
@igrigorik There is bytesJS to correspond to what I beleive is the total amount of js downloaded for a page.
If you graph the data in a scatter plot you can find a positive trend line. I beleive there is alot of noice from sites that are infinte scroll with lots of images.
Do you have a suggestion on a quiery that might find a trend?
Throwing the data into R gives the following (after eliminating some outliers):
Pearson correlation between the two: 0.3878254 … which I guess qualifies as a “moderate positive” relationship. Intuitively, this makes sense, but I wouldn’t draw too many conclusions from it.
I agree from this data nothing can be concluded.