php - how to rate or rank votes -

i'm sorry if i'm wrong question want idea...i want have , idea of ranking algorithm include time submit there votes.

nice question!

okay lets bring on!

first of 1 thing cannot when calculating ratings bayesianaverage

you ran read on simplified takes care of following:

• entries little votes not true mean of votes have componentn of mean rating throughout dataset. example on imdb default rating somewhere @ 6.4. film 2 votes 10 stars each may still have between 6 , 7. more votes more meaning become alltogether , rating "pulled away" default. imdb implements minimum number of votes movies show in listings.

another thing find confusing is: why time of vote important? didn't mean time of entry voted on? in our movies example released movies more important?

but anyway! in both cases results achieved applying logarithmic functions.

for our movie example movies relevance multiplied

1 + 1/sqrt(1 + current_year - release_year ) 

so 1 socket rating every movie gets. movie teh current year have boost of 100% (200% relevance) above return true. last year 170%, 2 years old 157% , son on.

but difference of movie 1954 or 1963 far not great.

so remember:

• everything use in calculations. linear? may distort ratings? relations throughout dataset sane?

if want have recent votes cast more can exact same way weight votes. makes sense if want recent voted stuff "warmed up"... because hot , discussed in community example.

that beeing said remains hard work. lot of playing around etc.

let me give 1 last idea.

at company work calculate relevance movies.

we have config array store "weighting" of several factors in final relevance.

it looks this:

        $weights = array(                 "year" => 2, // release year                 "rating" =>13, // rating 0-100                 "cover" => 4,  // cover available?                 "shortdescription" => 4, // short descr available?                 "trailer" => 3, // trailer available?                 "imdbpr" => 13, // google pagerank of imdb site         ); 

then calculate value between 0 , 1 every metric. there different methods. let me show example of our rating (which aggregated rating of several platforms crawl , have different weightings themsevles etc.)

        $yeardiff = $data["year"] - date('y');         //year         if (!$data["year"]){                 $values['year'] = 0;         } else if($yeardiff==0) {                    $values['year'] = 1;         } else if($yeardiff < 3) {                 $values['year'] = 0.8;         } else if($yeardiff < 10) {                    $values['year'] = 0.6;         } else {                 $values['year'] = 1/sqrt(abs($yeardiff));         } 

so see hardcoded "age intervals" , relyed on sqrt function older movies. in fact difference there minimal sqrt example here poor. mathematical functions useful!

you can, example, use periodic functions sinus curves etc calculate seasonal relevance! example year has range 0-1 can use sinus function weight summer hits / winter hits / autumn hits current time of year!

one last example imdb pagerank. hardcoded there 10 different values possible , not distributed in statistical homogenous way (pagerank 1 or 2 worse none):

        if($imdbpr >= 7) {                  $values['imdbpr'] = 1;         } else if($imdbpr >= 6) {                 $values['imdbpr'] = 0.9;         } else if($imdbpr >= 5) {                 $values['imdbpr'] = 0.8;         } else if($imdbpr >= 4) {                 $values['imdbpr'] = 0.6;         } else if($imdbpr >= 3) {                 $values['imdbpr'] = 0.5;         } else if($imdbpr >= 2) {                 $values['imdbpr'] = 0.3;         } else if($imdbpr >= 1) {                 $values['imdbpr'] = 0.1;         } else if($imdbpr >= 0) {                 $values['imdbpr'] = 0.0;         } else {                 $values['imdbpr'] = 0.4; // no pagerank available. new         } 

then sum this:

        foreach($values $field=>$value) {                 $malus += ($value*$weights[$field]) / array_sum($weights);         } 

this may not exact answer question bit more , broadly, hope pointed in right direction , gave points thoughts can pick up!

have fun , success application!