Rearranging the sequence

ES6, 118 81 78 bytes

n=>[...[...Array(n).keys()].join``].map(c=>r/=(o[c]=-~o[c])/i++,o=[],i=r=1)&&r

Someone's bound to tell me there's a shorter way of concatenating the numbers up to n.

Saved a cool 37 bytes by taking @edc65's idea and running it on steroids. (Save an extra byte by using '|' instead of && but that limits the result to 31 bits.)

Edit: Saved 3 further bytes again thanks to @edc65.

APL (Dyalog Extended), 13 bytes

×/2!/+\⎕D⍧⍕⍳⎕

Try it online!

A full program. Uses ⎕IO←0.

How it works

×/2!/+\⎕D⍧⍕⍳⎕
            ⎕  ⍝ Take input from stdin (N)
           ⍳   ⍝ Generate range 0..N-1
          ⍕    ⍝ Stringify the entire array (S)
               ⍝ (The result has spaces between items)
       ⎕D      ⍝ The character array '0123456789'
         ⍧     ⍝ Count occurrences of each digit in S
×/2!/+\        ⍝ Calculate multinomial:
     +\        ⍝   Cumulative sum
  2!/          ⍝   Binomial of consecutive pairs
×/             ⍝   Product

The multinomial calculation comes from the following fact:

$$ \frac{(a_1 + a_2 + \cdots + a_n)!}{a_1! a_2! \cdots a_n!} = \frac{(a_1 + a_2)!}{a_1! a_2!} \times \frac{(a_1 + a_2 + \cdots + a_n)!}{(a_1 + a_2)! a_3! \cdots a_n!} $$

$$ = \frac{(a_1 + a_2)!}{a_1! a_2!} \times \frac{(a_1 + a_2 + a_3)!}{(a_1 + a_2)! a_3!} \times \frac{(a_1 + a_2 + \cdots + a_n)!}{(a_1 + a_2 + a_3)! \cdots a_n!} $$

$$ = \cdots = \binom{a_1 + a_2}{a_1} \binom{a_1 + a_2 + a_3}{a_1 + a_2} \cdots \binom{a_1 + \cdots + a_n}{a_1 + \cdots + a_{n-1}} $$

MATL, 21 bytes

:qV0h!4Y2=sts:pw"@:p/

Try it online!

Explanation

:q     % implicitly input N, and generate vector [0, 1, ..., N-1]
V      % convert to string representation of numbers. Contains spaces,
       % but no matter. Only characters '0', ..., '9' will be counted
0h     % append 0 character (not '0'), so that string is never empty
!      % convert string to column char array
4Y2    % string '0123456789' (row char array)
=      % test all pairs for equality
s      % sum each column. For N = 12 this gives [2, 4, 1, 1, ..., 1]
t      % duplicate
s      % compute sum. For N = 12 this gives 14
:p     % factorial
w      % swap. Now vector [2, 4, 1, 1, ..., 1] is on top
"      % for each number in that vector
  @:p  %   factorial
  /    %   divide
       % implicitly end loop
       % implicitly display