ML Wiki
Machine Learning Wiki - A collection of ML concepts, algorithms, and resources.

Jefferson's Method

Jefferson’s Method

This is a Parliamentary Allocation method.

The task is:

  • given
    • $p_i$ - the number of voters in favor of party $i$
    • $N$ - total number of parties, $i \in { 1, 2, …, N} \equiv P$
    • $n$ - total number of voters
    • the quota of $i$ is $q_i = S \cdot \cfrac{p_i}{n}$. Note that $q_i$ is a read number, not integer
  • allocate $S$ seats in parliament
    • $(s_1, …, s_N)$ s.t. $\sum s_i = S$
    • $s_i$ must be an integer

The main idea of this method is to satisfy the following constraint:

  • $s_i \ne 0, \cfrac{p_i}{s_i} \geqslant \cfrac{p_j}{s_j + 1}$

If this constraint is not respected, we have:

  • $\cfrac{p_i}{s_i} < \cfrac{p_j}{s_j + 1}$
  • but party $P_j$ won’t be happy about it: they may need more people to allocate one seat than $P_i$

So we give a place to a party $i$ with maximal $\cfrac{p_i}{\lfloor s_i \rfloor + 1}$ score

Meaning:

  • suppose $S=10, p_1 = 6373, q_1 = 6.4$
  • allocate $s_1 = 6$ seats to $P_1$
  • $\cfrac{p_1}{s_1 + 1}$ means “to get one additional seat they need 910 people in the worst case”

Example 1

$S = 10$

| | $p_i$ | $q_i$ | $\lfloor q_i \rfloor$ | $s_i$ | $\cfrac{p_i}{\lfloor s_i \rfloor + 1}$ | $P_1$ | 6373 | 6.373 | 6 | 6 | 910.42 || $P_2$ | 2505 | 2.505 | 2 | 2 | 835 || $P_3$ | 602 | 0.602 | 0 | 0 | 602 || $P_4$ | 520 | 0.520 | 0 | 0 | 502 || | | | 8 | 8 | | $P_1$ has the highest $\cfrac{p_1}{\lfloor s_1 \rfloor + 1}$ score

  • so allocating additional seat to them
  • note that we need to re-calculate the value $\cfrac{p_i}{\lfloor s_i \rfloor + 1}$ for $P_1$ after we allocate the seat to them

| | $p_i$ | $q_i$ | $\lfloor q_i \rfloor$ | $s_i $ | $\cfrac{p_i}{\lfloor s_i \rfloor + 1}$ | $P_1$ | 6373 | 6.373 | 6 | 7 | 796 || $P_2$ | 2505 | 2.505 | 2 | 2 | 835 || $P_3$ | 602 | 0.602 | 0 | 0 | 602 || $P_4$ | 520 | 0.520 | 0 | 0 | 502 || | | | 8 | 9 | | $P_2$ has the highest score now

  • so allocating the 10th seat to them

Example 2

$S = 10$

| | $p_i$ | $q_i$ | $\lfloor q_i \rfloor$ | $s_i$ | $\cfrac{p_i}{\lfloor s_i \rfloor + 1}$ | $P_1$ | 6373 | 6.4 | 6 | 6 | 910.42 || $P_2$ | 2505 | 2.505 | 2 | 2 | 768.33 || $P_3$ | 702 | 0.602 | 0 | 0 | 702 || $P_4$ | 620 | 0.520 | 0 | 0 | 620 || | | | 8 | 8 | | Give the seat to $P_1$, recalculate the score:

| | $p_i$ | $q_i$ | $\lfloor q_i \rfloor$ | $s_i$ | $\cfrac{p_i}{\lfloor s_i \rfloor + 1}$ | $P_1$ | 6373 | 6.4 | 6 | 7 | 796.6 || $P_2$ | 2505 | 2.505 | 2 | 2 | 768.33 || $P_3$ | 702 | 0.602 | 0 | 0 | 702 || $P_4$ | 620 | 0.520 | 0 | 0 | 620 || | | | 8 | 9 | | Again allocate the seat to $P_1$

| | $p_i$ | $q_i$ | $\lfloor q_i \rfloor$ | $s_i$ | $P_1$ | 6373 | 6.4 | 6 | 7 || $P_2$ | 2505 | 2.505 | 2 | 2 || $P_3$ | 702 | 0.602 | 0 | 0 || $P_4$ | 620 | 0.520 | 0 | 0 || | | | 8 | 10 | This shows that the method is still not perfect.

Properties

Consistency

Show that

  • $\forall i, j \in P: p_i < p_j \Rightarrow s_i \leqslant s_j$

Solution

  • Jefferson Rule is: $\cfrac{p_i}{s_i} \geqslant \cfrac{p_j}{s_j + 1}$
  • or $\cfrac{p_i}{p_j} \geqslant \cfrac{s_i}{s_j + 1}$
  • $\cfrac{p_i}{p_j} < 1$ always (by the hypothesis $p_i < p_j$)
  • thus, $\cfrac{s_i}{s_j + 1} < 1$ or $s_i < s_j + 1$

Monotonicity

Also respected

  • http://www.math.colostate.edu/~spriggs/m130/apportionment2.pdf

See also

Sources