Question

I have a thousand recipes each having a tweet and facebook like counts. What i want to do is to create an overall rating out of 100 based off these two scores (and perhaps other social network counts too).

Assuming both facebook and twitter are equally weighted, how can i go about this.

one way to do this for any given network would be somethign like this

this_recipes_facebook_count / max_facebook_count_in_db * 100.0

and average it with the twitter result.

However what happens if there is a recipe with a freakish high score? It unfairly punishes other recipes with lower yet still relatively high scores.

I feel i need to take standard deviation into acccount, perhaps some dampening function...but its been 14 years since i took stats in highschool.

Can anyone help? Id prefer simple over complex as it is only recipe ratings after all.

Was it helpful?

Solution

Instead of linearly increasing the popularity count you might do something like this: (1-p^x)

Where p is a pre-selected value (say 0.99) and x is the number of mentions.

Initially increase in mentions is going to speed up the score a lot. But after sometime the effect becomes smaller and smaller.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top