site_ranking_v1

//	A few notes:
//
//	This is in pseudocode, and I'm using ** as the exponentiation function
//	here. I'm only including a subset of the data you have available, the stuff
//	that I think is probably most interesting for ranking. Most of the
//	functions are pretty straightforward, you should be able to add your own
//	without too much trouble.
//
//	Most of these are exponential functions that define a rate raised to
//	the power of some input value. This will produce your standard exponential
//	graph. If you want to move the graph along the X axis (i.e., make your
//	peak values happen sooner or later), then add to or subtract from the
//	exponent. If you want to move the graph up and down along the Y axis
//	(i.e., define greater or smaller minimum values), then simply add to or
//	subtract from the result.

//	Let's get started.

//	Tunable knobs.
//	Each tunable has a description in the function that uses it below.

ipb_rate = 1.2;
dslu_rate = 1.4;
dsc_rate = 1.01;
avg_hits = 2000;
sh_rate = .3;
shltp_rate = .4;
email_penalty = .8;


//	Site's overall starting score.
//
//	This just needs to be any nonzero positive value. Larger starting
//	values will result in a larger range of ranking scores.

score = 1;


//	How big the index.html is, in bytes.
//	ipb_rate is your tuning value here. Set it to any non-zero positive
//	value. Higher values make the size of the index.html file more
//	important.
//
//	I've divided index_page_bytes here by 1024 to penalize sites that
//	have index pages smaller than 1K, and to keep the exponent from
//	getting too ridiculously large.

score = score * (ipb_rate ** (index_page_bytes/1024));


//	When site was updated
//	dslu_rate is your tuning value here. days_since_last_update is being
//	used as a denominator to invert the result of the function (i.e.,
//	the more days since it was last updated, the smaller the resulting
//	score is). Sites that have been updated less than 5 days ago get a boost.

score = score * (dslu_rate ** (5/days_since_last_update));


//	When site was created
//	dsc_rate is your tuning value here. Use a small dsc_rate value to give
//	older sites a nice boost in rankings without penalizing newer sites.

score = score * (dsc_rate ** days_since_creation);


//	How many hits the site has gotten
//	sh_rate is your tuning value here. I'm including another value here,
//	avg_hits, to establish a sort of baseline -- sites with higher than
//	average hits get a boost, sites with smaller than average hits get a
//	small penalty. If you don't know your avg_hits value, just treat it as
//	a tunable parameter and pick a value that seems reasonable.
//
//	I've also switched the order of the base and the exponent here, since
//	we should expect that (site_hits / avg_hits) could get large.

score = score * ((site_hits / avg_hits) ** sh_rate);


//	How many hits the site has gotten in the last (time period)
//	shltp_rate (site-hits-last-time-period rate) is your tuning value here.
//	I think this would be a good additional parameter, if you can do it.
//	This will help to measure the trending popularity of newer sites,
//	and should be given a good boost.

score = score * (site_hits_last_24h ** shltp_rate);


//	Whether the site provided an email on creation or not
//
//	I assume here you want to heavily penalize sites that didn't supply an
//	email address. email_penalty can be any value from .01 to .99; larger
//	values result in a bigger penalty for not having an email address.
//
//	Let's say your email_penalty is .8; if a site has an email address, then
//	its score will be multiplied by 1.2, a relatively slight upward adjustment.
//	But, if your site doesn't have an email address, its score will be
//	multiplied by 0.2 -- a severe downward adjustment.

score = score * ((site_has_email_addr ? 1 : 0) + (1 - email_penalty));


//	Done.