H
H
hugovox2011-12-13 18:39:56
Ratings
hugovox, 2011-12-13 18:39:56

Need help with a mathematical formula to calculate the rating

Help to make a formula for rating. The task is to build a rating of sites based on the following indicators:

1. Site
traffic The Alexa indicator (Traffic Rank in RU) is taken as the basis, the indicator can be from 1 to 100,000, the higher the indicator, the worse.

2. Brand awareness
Two indicators are taken as a basis:
- the number of requests in Yandex, the indicator can be from 1 to 200,000, the higher the indicator, the better
- mentions in blogs, the indicator can be from 1 to 200,000, the higher the indicator, the better better

3.
Feedback score Based on user feedback score, the indicator can be from 1 to 10, the higher the score, the better.

The weight of each indicator in the total:
Site traffic - 35%
Brand awareness - 35%
Customer feedback score - 30%

Need help in compiling a mathematical formula for calculating the rating so that all indicators can be reduced to one figure. Ready for a reward. Thank you for your attention.

PS if there are errors in the methodology, I will be glad to comments

Answer the question

In order to leave comments, you need to log in

3 answer(s)
B
Bodigrim, 2011-12-14
@Bodigrim

Given the spread of the scales in 1 and 2, I would use a logarithmic renormalization for them. For example, it is clear that the difference between sites with 10 and 20 blog mentions is much larger than between sites with 1000 and 1010 blog mentions. The linear scales proposed above will not capture this.
Then it will turn out something like this (in the notation from the first comment):
rate = 0.35*(1-log(a)/log(100000)) + 0.35*(log(b1)+log(b2))/log(200000) /2 + 0.30*(s-1)/9.
Here rate is 0 at worst, 1 at best.

T
TheHorse, 2011-12-13
@TheHorse

a - Attendance
b1, b2 - Fame
c - Score
total = round(((100000 - a) / 99999 * 0.35 + (b1 + b2 - 2) / 399998 * 0.35 + (c - 1) / 9 * 0.3) * r )
where r is any number. For example, when r = 100; The rating will be from 0 to 100.

R
Roman Gogolev, 2011-12-14
@romka777

rating = 0.35 * (100000 - k1) / 100000 + 0.35 * (k1 / 200000 + k2 / 200000) + 0.30 * k3 / 10

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question