Power curve adjustment for social media queries

Question

Twitter recently announced that you can get close to the rank of any given Twitter user with high accuracy by entering your blind count in the following formula:

exp ($ a + $ b * log (follower_count))

where $ a = 21 and $ b = -1.1

This, obviously, is much more effective than sorting the entire list of users by the followers account for a given user.

If you have a similar dataset from another social site, how could you get the values ​​for $ a and $ b to match this dataset? In principle, a list of frequencies whose distribution is assumed to be power-law.

Answer 1

:

y = exp(a + b.log(x))

:

log(y) = a + b.log(x)

, , , , a b.

. , , ?

Answer 2

Microsoft Excel "Solver". Excel, . "" "" Excel .

:

• . A ()

• B, .

• , , B.

• C ( , 1, 2, 3 ..)

• 21 D1 -1.1 E1. Twitter $A $B. . , , .

• D2 ​​: = exp ($ E $1 + $F $1 * log (B2))

• D2 .

• E2 (.. ). , = sqrt (c2 * c2 + d2 * d2). , 0.

• E2 .

• , E, . , , 10 000 . E10001 = (e2: e10000).

• Data "Solver". Excel. "" .

• ( ) , Solver. , D1 E1, , E10001 ( ) .