0-1 Maximal Phase Counter

Question

Consider an array of bits, say

1 1 1 0 0 0 0 1 0 0 1 0 1 1 1 1 1 0 1 0

We call a contiguous subarray of length ≥ 5 a phase if at least 85% of the bits are the same and the first/last bits are both equal to the majority bit. Furthermore, we call a phase maximal if it is not a strict subarray of some other phase.

Here are the maximal phases of the above example:

1 1 1 0 0 0 0 1 0 0 1 0 1 1 1 1 1 0 1 0
      -------------
                    -------------
                        -------------

As you can see, there are 3 maximal phases. On the other hand, this

1 1 1 0 0 0 0 1 0 0 1 0 1 1 1 1 1 0 1 0
                        ---------

is not a maximal phase as it is a strict subarray of at least one other phase.

The challenge

Input is a sequence of ≥ 5 bits via STDIN, command line or function argument. The bits may come in as a string or an array.

You are to output a single integer, the number of maximal phases for the array, either printed via STDOUT or returned from a function.

Scoring

This is code-golf so the program in the fewest bytes wins.

Test cases

0 1 0 1 0 -> 0
0 0 0 0 0 -> 1
0 0 0 0 1 0 1 1 1 1 -> 0
0 0 0 0 0 1 0 1 1 1 1 1 -> 2
1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 -> 1
0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 -> 2
0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 -> 1
0 1 0 1 0 0 1 0 1 0 1 0 0 0 1 1 1 1 0 1 0 0 1 1 0 0 0 1 1 0 -> 0
1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 0 0 1 0 0 0 0 0 0 0 0 1 1 0 1 -> 4
0 1 1 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 0 1 0 0 0 0 0 -> 5

Here's the explanation for the last case:

0 1 1 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 0 1 0 0 0 0 0
---------------------------
      -------------------------
                            -----------------
                                -----------------
                                              -------------

^{Fun fact: This challenge came about from a data mining problem with the goal of detecting change in temporal data.}

Answer 1

Python 2, 149 bytes

a=input()
l=len(a)
n=p=0
for i in range(l):
 for j in range(l-1,i+3,-1):
  if(j>p)>(.15<sum(a[i:j+1])/(j+1.-i)+a[i]+a[j]<2.85):n+=1;p=j;break
print n

The first loop scans across the array from left to right. Each bit, indexed by i, is checked to see if it could be the first bit in a maximal phase.

This is done by the inner loop, which scans from right to left. If the subarray between i and j is a phase, we increase the counter and move on. Otherwise, we keep going until the subarray becomes too small or j reaches the end of the previous maximal phase.

1 1 1 0 0 0 0 1 0 0 1 0 1 1 1 1 1 0 1 0
i ->                               <- j

Example:

$ python phase.py
[1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 0]
3

Answer 2

Python 2, 144

Enter input in the form [0,1,0,1,0].

a=input()
o=[2];i=-1
while a[i:]:
 j=len(a);i+=1
 while j>i+4:o+=sum(j>max(o)>x==a[i]==a[j-1]for x in a[i:j])*20/(j-i)/17*[j];j-=1
print~-len(o)

Subsequences are checked with ordering by increasing initial element, then decreasing length. In this way, it is known that a new subsequence is not a subsequence of a previous subsequence iff the index of its last element is greater than any index of a previously found sequence's last element.

Answer 3

Dyalog APL, 62 characters

{≢∪0~⍨+/∨⍀∨\⌽∘.{((⊃=⊃∘⌽)∧(.85≤(+/⊢=⊃)÷≢)∧5≤≢)(⍺-1)↓⍵↑a}⍨⍳⍴a←⍵}

Similar to Zgarb's solution but golfed a little further.

Answer 4

Dyalog APL, 86 bytes*

{+/∨/¨∪↓∨⍀∨\{⊃({(.5>|k-⍵)∧.35≤|.5-⍵}(+/÷⍴)⍵)∧(5≤⍴⍵)∧(⊃⌽⍵)=k←⊃⍵}¨⌽∘.{(⍺-1)↓⍵↑t}⍨⍳⍴t←⍵}

Try it here. Usage:

   f ← {+/∨/¨∪↓∨⍀∨\{⊃({(.5>|k-⍵)∧.35≤|.5-⍵}(+/÷⍴)⍵)∧(5≤⍴⍵)∧(⊃⌽⍵)=k←⊃⍵}¨⌽∘.{(⍺-1)↓⍵↑t}⍨⍳⍴t←⍵}
   f 0 0 0 0 0 1 0 1 1 1 1 1
2

This can probably be golfed quite a bit, especially the middle part, where the phase condition is checked.

Explanation

I first collect the substrings of the input vector into a matrix, where the upper left corner contains the whole input, using ⌽∘.{(⍺-1)↓⍵↑t}⍨⍳⍴t←⍵. For the input 0 0 0 0 0 1 0, this matrix is

┌───────────────┬─────────────┬───────────┬─────────┬───────┬─────┬───┬─┐
│1 0 0 0 0 0 1 0│1 0 0 0 0 0 1│1 0 0 0 0 0│1 0 0 0 0│1 0 0 0│1 0 0│1 0│1│
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│0 0 0 0 0 1 0  │0 0 0 0 0 1  │0 0 0 0 0  │0 0 0 0  │0 0 0  │0 0  │0  │ │
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│0 0 0 0 1 0    │0 0 0 0 1    │0 0 0 0    │0 0 0    │0 0    │0    │   │ │
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│0 0 0 1 0      │0 0 0 1      │0 0 0      │0 0      │0      │     │   │ │
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│0 0 1 0        │0 0 1        │0 0        │0        │       │     │   │ │
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│0 1 0          │0 1          │0          │         │       │     │   │ │
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│1 0            │1            │           │         │       │     │   │ │
├───────────────┼─────────────┼───────────┼─────────┼───────┼─────┼───┼─┤
│0              │             │           │         │       │     │   │ │
└───────────────┴─────────────┴───────────┴─────────┴───────┴─────┴───┴─┘

Then I map the condition of being a phase over it, resulting in the 0-1-matrix

0 0 0 0 0 0 0 0
1 0 1 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0

To get the number of maximal phases, I flood the 1's to the right and down using ∨⍀∨\,

0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1

collect the unique rows with ∪↓,

┌───────────────┬───────────────┐
│0 0 0 0 0 0 0 0│1 1 1 1 1 1 1 1│
└───────────────┴───────────────┘

and count those that contain at least one 1 using +/∨/¨.

^{*There exists a standard 1-byte encoding for APL.}

Answer 5

Clojure, 302

(defn p[v l](if(or(<(count v)5)(= 0 l))nil(if((fn[v](let[f(first v)c(apply + v)o(count v)r(/ c o)t(+ f f r)](and(= f(last v))(or(> t 2.85)(< t 0.15)))))v)0(let[r(p(vec(drop-last v))(dec l))](if r(+ r 1)r)))))(defn s[v l c](if(empty? v)c(let[n(p v l)](if n(s(vec(rest v))n(inc c))(s(vec(rest v))l c)))))

and the slightly ungolfed version

(defn is-phase [vector]
  (let [f (first vector)
        c (apply + vector)
        o (count vector)
        r (/ c o)
        t (+ f f r)]
    (and (= f (last vector))
         (or (> t 2.85) (< t 0.15)))))
(defn phase-index [vector last]
  (if (or (<(count vector)5)(= 0 last)) nil
    (if (is-phase vector) 0
      (let [r (phase-index (vec(drop-last vector)) (dec last))]
        (if r (+ r 1) r)))))
(defn phase-count [vector last count]
  (if (empty? vector) count
    (let [n (phase-index vector last)]
         (if n (phase-count (vec(rest vector)) n (inc count))
             (phase-count (vec(rest vector)) last count)))))

callable like this: (s [0 1 0 1 0] 10 0). It requires a few extra arguments, but I could get rid of those with an extra 20 characters.

Answer 6

JavaScript (ES6) 141

@grc's algorithm ported to JavaScript
Input can be a string or an array

F=b=>
  (l=>{
    for(c=e=i=0;i<l;++i)
      for(j=l;j>i+4&j>e;--j)
        (k=0,[for(d of b.slice(i,j))k+=d==b[i]],k<(j-i)*.85)|b[i]-b[j-1]||(++c,e=j)
  })(b.length)|c

Test In FireFox / FireBug console

;['01010', '00000', '0000101111',
'000001011111', '100000000000010',
'0000010000010000010', '00000100000100000100',
'010100101010001111010011000110',
'111110000011111001000000001101',
'011000000000001011111110100000'].forEach(t => console.log(t,F(t)))

Output

01010 0
00000 1
0000101111 0
000001011111 2
100000000000010 1
0000010000010000010 2
00000100000100000100 1
010100101010001111010011000110 0
111110000011111001000000001101 4
011000000000001011111110100000 5

Answer 7

CJam, 110 103 bytes

Pretttttty long. Can be golfed a lot.

q~_,,\f>{_,),5>\f<{:X)\0==X1b_X,.85*<!\.15X,*>!X0=!*\X0=*+&},:,W>U):U+}%{,(},_{{_W=IW=>\1bI1b>!&!},}fI,

Input is like

[0 1 1 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 0 1 0 0 0 0 0]

Output is the number of phases.

Try it online here

Answer 8

JavaScript (ECMAScript 6), 148 139 Bytes

f=(s,l=0,e=0,p=0)=>{for(n=s.length,o=[j=0,y=0],i=l;i<n;++j>4&x==s[l]&i>e&c>=.85‌​*j&&(e=i,y=1))c=++o[x=s[i++]];return l-n?f(s,l+1,e,p+y):p}

Recurses through the array and starts iteration at the last recursion index. Argument can be either an array or string.

f('011000000000001011111110100000'); //5

Answer 9

Wolfram - 131

{x_, X___}⊕{Y__, x_, y___}/;MemberQ[t={x, X, Y, x}, 1-x] && t~Count~x > .85 Length@t := 
  1 + {X, Y, x}⊕{y} 
{_, X___}⊕y_ := {X}⊕y
{}⊕{y_, Y__} := {y}⊕{Y}
_⊕_ := 0

Example

{}⊕{1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 0}
> 3
{}⊕{0,1,1,0,0,0,0,0,0,0,0,0,0,0,1,0,1,1,1,1,1,1,1,0,1,0,0,0,0,0}
> 5

Answer 10

Java: 771 bytes

import java.util.*;public class A{static int[]a;static class b{int c,d,e,f,g,h;b(int i,int j){this.c=i;this.d=j;this.h=j-i+1;this.e=k();this.f=this.h-this.e;this.g=e>f?1:0;}
boolean l(b n){return this.c>=n.c&&this.d<=n.d;}
int k(){int o=0;for(int i=c;i<=d;i++){if(a[i]==1){o++;}}
return o;}
public boolean equals(Object o){b x=(b)o;return x.c==this.c&&x.d==this.d;}
float p(){if(g==0){return(float)f/h;}else{return(float)e/h;}}
boolean q(){float r=p();return a[c]==a[d]&&a[d]==g&&r>=0.85F;}}
static int s(int[]t){a=t;List<b>u=new ArrayList<>();for(int v=0;v<t.length-4;v++){int x=v+4;while(x<t.length){b y=new b(v,x);if(y.q()){u.add(y);}
x++;}}
List<b>a=new ArrayList<>();for(b c:u){for(b d:u){if(!c.equals(d)&&c.l(d)){a.add(c);break;}}}
u.removeAll(a);return u.size();}}

run by calling method s(int[] input)