ericnormand/00 quartiles.md

## 00 quartiles.md

      
    Raw
  

              00 quartiles.md
            
          
    calculate quartiles
It's often useful to know the quartiles of a dataset. The quartiles define the lowest quarter, lowest half, and lowest three quarters data points. Those are typically called q1, q2, and q3. I also like to include q0 and q4, which are the min and max.
q2 is just the median of the dataset. q1 is the median of the lower half of the dataset. And q3 is the median of the upper half of the dataset.
Write a function which calculates the quartiles of a sequence of numbers. You will need to be able to calculate the median (which you can look up easily).

  
## Eric Normand.clj
(defn average [a b]
  (/ (+ a b) 2))

(defn median [sorted-vec]
  (let [h (quot (count sorted-vec) 2)]
    (if (even? (count sorted-vec))
      (average (get sorted-vec (dec h)) (get sorted-vec h))
      (get sorted-vec h))))

(defn halves [sorted-vec]
  (let [h (quot (count sorted-vec) 2)]
    (if (even? (count sorted-vec))
      [(subvec sorted-vec 0 h)
       (subvec sorted-vec h)]
      [(subvec sorted-vec 0 (inc h))
       (subvec sorted-vec h)])))

(defn quartiles
  "Calculate the quartiles with Method 2 from https://en.wikipedia.org/wiki/Quartile"
  [numbers]
  (let [sorted-vec (vec (sort numbers))
        [h1 h2] (halves sorted-vec)]
    {:q0 (first sorted-vec)
     :q1 (median h1)
     :q2 (median sorted-vec)
     :q3 (median h2)
     :q4 (last sorted-vec)}))

## Ilya Bernshteyn.clj
(ns clj-challenge.quartiles)

(defn median
  "Returns the median value of a collection. The count `cnt` can be passed in to
  save time."
  ([coll]
   (median coll (count coll)))
  ([coll cnt]
   (cond
     (odd? cnt) (nth coll (/ cnt 2))
     (even? cnt) (/ (+ (nth coll (/ cnt 2))
                       (nth coll (/ (- cnt 1) 2)))
                    2))))

(defn halves
  "Returns the ((left) (right)) halves of a collection. The count `cnt` can be passed
  in to save time."
  ([coll]
   (halves coll (count coll)))
  ([coll cnt]
   (cond
     (odd? cnt) (let [[left right] (split-at (/ cnt 2) coll)]
                  (list (butlast left) right))
     (even? cnt) (split-at (/ cnt 2) coll))))

(defn quartiles
  "Returns the quartiles of a collection, like this:
    {:q0 1  ;; the min
     :q1 2  ;; 25th percentile
     :q2 3  ;; the median
     :q3 4  ;; 75th percentile
     :q4 5} ;; the max"
  [coll]
  (let [sorted (sort coll)
        cnt (count sorted)
        cnt-half (quot cnt 2)
        [left right] (halves sorted cnt)]
    {:q0 (first sorted)
     :q1 (median left cnt-half)
     :q2 (median sorted cnt)
     :q3 (median right cnt-half)
     :q4 (last sorted)}))

;; REPL expressions to try things out
(comment
  ;; Sample data
  (def even [1 10 2 9 3 8 4 7 5 6])
  (def odd [1 10 2 9 3 8 4 7 5 6 11])
  (def big (take 1000000 (map rand-int (repeat 1000000))))
  ;; Get median, halves
  (median even)
  (median odd)
  (halves even)
  (halves odd)
  ;; Even number of elements
  (quartiles even)
  ;; Odd number of elements
  (quartiles odd)
  ;; Big random numbers
  (time (quartiles big)))

## Jos van Bakel.clj
(ns scratchpad.pftv.335
  (:require [clojure.string :as str]
            [clojure.test :refer [deftest is testing run-tests]]))

;; uses "Method 1" from https://en.wikipedia.org/wiki/Quartile#Computing_methods
;; assumes v is a sorted vec
(defn median* [v]
  (if (not (empty? v))
    (let [len (count v)
          mid (quot len 2)]
      (if (even? len)
        [(/ (+ (nth v (dec mid)) (nth v mid)) 2)
         (subvec v 0 mid)
         (subvec v mid)]
        [(nth v mid)
         (subvec v 0 mid)
         (subvec v (inc mid))]))))

(defn median [lst]
  (first (median* (vec (sort lst)))))

(defn quartiles [lst]
  (let [v (vec (sort lst))
        q0 (first v)
        [q2 left right] (median* v)
        [q1 & _] (median* left)
        [q3 & _] (median* right)
        q4 (last v)]
    (if q2
      [q0 q1 q2 q3 q4])))

(deftest test-median
  (testing "All cases of median"
    (is (nil? (median [])))
    (is (= (median [1]) 1))
    (is (= (median [1 2]) 3/2))
    (is (= (median [1 2 3]) 2)))
  (testing "Median of Wikipedia example"
    (is (= (median [6 7 15 36 39 40 41 42 43 47 49]) 40))))

(deftest test-quartiles
  (testing "All cases of quartiles"
    (is (nil? (quartiles [])))
    (is (= (quartiles [1]) [1 nil 1 nil 1]))
    (is (= (quartiles [1 2]) [1 1 3/2 2 2]))
    (is (= (quartiles [1 2 3 4 5 6 7]) [1 2 4 6 7])))
  (testing "Quartiles of Wikipedia example"
    (is (= (quartiles [6 7 15 36 39 40 41 42 43 47 49]) [6 15 40 43 49]))))

(run-tests)

## Mark Champine.clj
(defn halve [s]
  (split-at (int (/ (count s) 2)) (sort s)))

(defn median [s]
  (let [[lh [fl & _]] (halve s)]
    (if (even? (count s)) (/ (+ (last lh) fl) 2.0) fl)))

; uses quartile method 1 from https://en.wikipedia.org/wiki/Quartile
(defn quartiles [s]
  (let [ss (sort s)
        [lh uh] (halve s)
        uhm (if (even? (count uh)) uh (rest uh))]
    {:q0 (first ss)
     :q1 (median lh)
     :q2 (median ss)
     :q3 (median uhm)
     :q4 (last ss)}))

;; Tests

;; check against sample data and answer from
;; https://www.statisticshowto.datasciencecentral.com/calculators/interquartile-range-calculator/
(quartiles '(12, 13, 15, 18, 19, 22, 88, 89, 90, 91, 92, 93, 95, 98, 99, 101, 101, 103, 105, 106, 107, 108, 109, 200, 201, 201, 203, 204, 215, 216, 217, 222, 223, 224, 225, 227, 229, 230, 232, 245, 246, 250, 258, 270, 271, 271, 272, 273))
;; {:q0 12, :q1 94.0, :q2 200.5, :q3 228.0, :q4 273}

(quartiles '(1 2 3 4 5))
;; {:q0 1, :q1 1.5, :q2 3, :q3 4.5, :q4 5}

(quartiles '(1 2 3 4 5 6))
;; {:q0 1, :q1 2, :q2 3.5, :q3 5.5, :q4 6}

(quartiles (range 1 100))
;; {:q0 1, :q1 25, :q2 50, :q3 74.5, :q4 99}

(defn genrndnseq [l n]
  (repeatedly l #(rand-int n)))

;; runs with increasing numbers of randoms 0-100
(quartiles (genrndnseq 10 100)).    ;; {:q0 8, :q1 40,   :q2 56.0, :q3 78.0, :q4 89}
(quartiles (genrndnseq 100 100))    ;; {:q0 0, :q1 33.5, :q2 56.5, :q3 82.5, :q4 98}
(quartiles (genrndnseq 1000 100))   ;; {:q0 0, :q1 25.0, :q2 49.0, :q3 74.0, :q4 100}
(quartiles (genrndnseq 10000 100))  ;; {:q0 0, :q1 25.0, :q2 51.0, :q3 75.0, :q4 100}
(quartiles (genrndnseq 100000 100)) ;; {:q0 0, :q1 25.0, :q2 50.0, :q3 75.0, :q4 100}

## Steve Miner.clj
(ns miner.quartiles)

;; https://purelyfunctional.tv/issues/purelyfunctional-tv-newsletter-335-idea-when-can-you-use-property-based-testing/

;; Quartiles
;;
;; q2 is just the median of the dataset. q1 is the median of the lower half of the
;; dataset. And q3 is the median of the upper half of the dataset.  I also like to include
;; q0 and q4, which are the min and max.


;; We have to decide if the median belongs to the lower or upper halves (or neither).
;; I'm using "Tukey's hinges" or what Wikipedia calls "Method 2".  If the original data has an even
;; count, split the data exactly in half.  If it has an odd count, include the median in
;; both halves.  This convention eliminates an awkward edge case for single element data sets.
;;
;; https://en.wikipedia.org/wiki/Quartile


(defn- median-sorted-vec [v]
  (let [cnt (count v)
        mid (quot cnt 2)]
    (if (odd? cnt)
      (v mid)
      (/ (+ (v (dec mid)) (v mid)) 2.0))))

(defn quartiles [coll]
  "Returns a vector of 5 elements based on the collection of numbers `coll`. Q0 is the
  minimum, Q1 is the median of the lower half, Q2 is the median, Q3 is the median of the
  upper half, and Q4 is the maximum.  Returns nil for the empty collection.  The halves are
  split according to the *Tukey's hinges* convention in which the median belongs to both
  halves if the dataset has an odd number of elements."
  (when (seq coll)
    (let [v (vec (sort coll))
          cnt (count v)
          lower (subvec v 0 (quot (inc cnt) 2))
          upper (subvec v (if (odd? cnt) (dec (count lower)) (count lower)))]
      [(v 0)
       (median-sorted-vec lower)
       (median-sorted-vec v)
       (median-sorted-vec upper)
       (v (dec cnt))])))

(defn smoke-test-quartiles []
  (assert (= (quartiles [6, 7, 15, 36, 39, 40, 41, 42, 43, 47, 49])
             [6 25.5 40 42.5 49]))
  (assert (= (quartiles [7, 15, 36, 39, 40, 41])
             [7 15 37.5 40 41]))
  (assert (= (quartiles (range 0.0 10.1 0.5))
             [0.0 2.5 5.0 7.5 10.0]))
  (assert (= (quartiles [13])
             [13 13 13 13 13]))
  (assert (= (quartiles (range 100001))
             [0 25000 50000 75000 100000]))
  (assert (nil? (quartiles [])))
  true)
	(defn average [a b]
	(/ (+ a b) 2))

	(defn median [sorted-vec]
	(let [h (quot (count sorted-vec) 2)]
	(if (even? (count sorted-vec))
	(average (get sorted-vec (dec h)) (get sorted-vec h))
	(get sorted-vec h))))

	(defn halves [sorted-vec]
	(let [h (quot (count sorted-vec) 2)]
	(if (even? (count sorted-vec))
	[(subvec sorted-vec 0 h)
	(subvec sorted-vec h)]
	[(subvec sorted-vec 0 (inc h))
	(subvec sorted-vec h)])))

	(defn quartiles
	"Calculate the quartiles with Method 2 from https://en.wikipedia.org/wiki/Quartile"
	[numbers]
	(let [sorted-vec (vec (sort numbers))
	[h1 h2] (halves sorted-vec)]
	{:q0 (first sorted-vec)
	:q1 (median h1)
	:q2 (median sorted-vec)
	:q3 (median h2)
	:q4 (last sorted-vec)}))
	(ns clj-challenge.quartiles)

	(defn median
	"Returns the median value of a collection. The count `cnt` can be passed in to
	save time."
	([coll]
	(median coll (count coll)))
	([coll cnt]
	(cond
	(odd? cnt) (nth coll (/ cnt 2))
	(even? cnt) (/ (+ (nth coll (/ cnt 2))
	(nth coll (/ (- cnt 1) 2)))
	2))))

	(defn halves
	"Returns the ((left) (right)) halves of a collection. The count `cnt` can be passed
	in to save time."
	([coll]
	(halves coll (count coll)))
	([coll cnt]
	(cond
	(odd? cnt) (let [[left right] (split-at (/ cnt 2) coll)]
	(list (butlast left) right))
	(even? cnt) (split-at (/ cnt 2) coll))))

	(defn quartiles
	"Returns the quartiles of a collection, like this:
	{:q0 1 ;; the min
	:q1 2 ;; 25th percentile
	:q2 3 ;; the median
	:q3 4 ;; 75th percentile
	:q4 5} ;; the max"
	[coll]
	(let [sorted (sort coll)
	cnt (count sorted)
	cnt-half (quot cnt 2)
	[left right] (halves sorted cnt)]
	{:q0 (first sorted)
	:q1 (median left cnt-half)
	:q2 (median sorted cnt)
	:q3 (median right cnt-half)
	:q4 (last sorted)}))

	;; REPL expressions to try things out
	(comment
	;; Sample data
	(def even [1 10 2 9 3 8 4 7 5 6])
	(def odd [1 10 2 9 3 8 4 7 5 6 11])
	(def big (take 1000000 (map rand-int (repeat 1000000))))
	;; Get median, halves
	(median even)
	(median odd)
	(halves even)
	(halves odd)
	;; Even number of elements
	(quartiles even)
	;; Odd number of elements
	(quartiles odd)
	;; Big random numbers
	(time (quartiles big)))
	(ns scratchpad.pftv.335
	(:require [clojure.string :as str]
	[clojure.test :refer [deftest is testing run-tests]]))

	;; uses "Method 1" from https://en.wikipedia.org/wiki/Quartile#Computing_methods
	;; assumes v is a sorted vec
	(defn median* [v]
	(if (not (empty? v))
	(let [len (count v)
	mid (quot len 2)]
	(if (even? len)
	[(/ (+ (nth v (dec mid)) (nth v mid)) 2)
	(subvec v 0 mid)
	(subvec v mid)]
	[(nth v mid)
	(subvec v 0 mid)
	(subvec v (inc mid))]))))

	(defn median [lst]
	(first (median* (vec (sort lst)))))

	(defn quartiles [lst]
	(let [v (vec (sort lst))
	q0 (first v)
	[q2 left right] (median* v)
	[q1 & _] (median* left)
	[q3 & _] (median* right)
	q4 (last v)]
	(if q2
	[q0 q1 q2 q3 q4])))

	(deftest test-median
	(testing "All cases of median"
	(is (nil? (median [])))
	(is (= (median [1]) 1))
	(is (= (median [1 2]) 3/2))
	(is (= (median [1 2 3]) 2)))
	(testing "Median of Wikipedia example"
	(is (= (median [6 7 15 36 39 40 41 42 43 47 49]) 40))))

	(deftest test-quartiles
	(testing "All cases of quartiles"
	(is (nil? (quartiles [])))
	(is (= (quartiles [1]) [1 nil 1 nil 1]))
	(is (= (quartiles [1 2]) [1 1 3/2 2 2]))
	(is (= (quartiles [1 2 3 4 5 6 7]) [1 2 4 6 7])))
	(testing "Quartiles of Wikipedia example"
	(is (= (quartiles [6 7 15 36 39 40 41 42 43 47 49]) [6 15 40 43 49]))))

	(run-tests)
	(defn halve [s]
	(split-at (int (/ (count s) 2)) (sort s)))

	(defn median [s]
	(let [[lh [fl & _]] (halve s)]
	(if (even? (count s)) (/ (+ (last lh) fl) 2.0) fl)))

	; uses quartile method 1 from https://en.wikipedia.org/wiki/Quartile
	(defn quartiles [s]
	(let [ss (sort s)
	[lh uh] (halve s)
	uhm (if (even? (count uh)) uh (rest uh))]
	{:q0 (first ss)
	:q1 (median lh)
	:q2 (median ss)
	:q3 (median uhm)
	:q4 (last ss)}))

	;; Tests

	;; check against sample data and answer from
	;; https://www.statisticshowto.datasciencecentral.com/calculators/interquartile-range-calculator/
	(quartiles '(12, 13, 15, 18, 19, 22, 88, 89, 90, 91, 92, 93, 95, 98, 99, 101, 101, 103, 105, 106, 107, 108, 109, 200, 201, 201, 203, 204, 215, 216, 217, 222, 223, 224, 225, 227, 229, 230, 232, 245, 246, 250, 258, 270, 271, 271, 272, 273))
	;; {:q0 12, :q1 94.0, :q2 200.5, :q3 228.0, :q4 273}

	(quartiles '(1 2 3 4 5))
	;; {:q0 1, :q1 1.5, :q2 3, :q3 4.5, :q4 5}

	(quartiles '(1 2 3 4 5 6))
	;; {:q0 1, :q1 2, :q2 3.5, :q3 5.5, :q4 6}

	(quartiles (range 1 100))
	;; {:q0 1, :q1 25, :q2 50, :q3 74.5, :q4 99}

	(defn genrndnseq [l n]
	(repeatedly l #(rand-int n)))

	;; runs with increasing numbers of randoms 0-100
	(quartiles (genrndnseq 10 100)). ;; {:q0 8, :q1 40, :q2 56.0, :q3 78.0, :q4 89}
	(quartiles (genrndnseq 100 100)) ;; {:q0 0, :q1 33.5, :q2 56.5, :q3 82.5, :q4 98}
	(quartiles (genrndnseq 1000 100)) ;; {:q0 0, :q1 25.0, :q2 49.0, :q3 74.0, :q4 100}
	(quartiles (genrndnseq 10000 100)) ;; {:q0 0, :q1 25.0, :q2 51.0, :q3 75.0, :q4 100}
	(quartiles (genrndnseq 100000 100)) ;; {:q0 0, :q1 25.0, :q2 50.0, :q3 75.0, :q4 100}
	(ns miner.quartiles)

	;; https://purelyfunctional.tv/issues/purelyfunctional-tv-newsletter-335-idea-when-can-you-use-property-based-testing/

	;; Quartiles
	;;
	;; q2 is just the median of the dataset. q1 is the median of the lower half of the
	;; dataset. And q3 is the median of the upper half of the dataset. I also like to include
	;; q0 and q4, which are the min and max.



	;; We have to decide if the median belongs to the lower or upper halves (or neither).
	;; I'm using "Tukey's hinges" or what Wikipedia calls "Method 2". If the original data has an even
	;; count, split the data exactly in half. If it has an odd count, include the median in
	;; both halves. This convention eliminates an awkward edge case for single element data sets.
	;;
	;; https://en.wikipedia.org/wiki/Quartile


	(defn- median-sorted-vec [v]
	(let [cnt (count v)
	mid (quot cnt 2)]
	(if (odd? cnt)
	(v mid)
	(/ (+ (v (dec mid)) (v mid)) 2.0))))

	(defn quartiles [coll]
	"Returns a vector of 5 elements based on the collection of numbers `coll`. Q0 is the
	minimum, Q1 is the median of the lower half, Q2 is the median, Q3 is the median of the
	upper half, and Q4 is the maximum. Returns nil for the empty collection. The halves are
	split according to the Tukey's hinges convention in which the median belongs to both
	halves if the dataset has an odd number of elements."
	(when (seq coll)
	(let [v (vec (sort coll))
	cnt (count v)
	lower (subvec v 0 (quot (inc cnt) 2))
	upper (subvec v (if (odd? cnt) (dec (count lower)) (count lower)))]
	[(v 0)
	(median-sorted-vec lower)
	(median-sorted-vec v)
	(median-sorted-vec upper)
	(v (dec cnt))])))

	(defn smoke-test-quartiles []
	(assert (= (quartiles [6, 7, 15, 36, 39, 40, 41, 42, 43, 47, 49])
	[6 25.5 40 42.5 49]))
	(assert (= (quartiles [7, 15, 36, 39, 40, 41])
	[7 15 37.5 40 41]))
	(assert (= (quartiles (range 0.0 10.1 0.5))
	[0.0 2.5 5.0 7.5 10.0]))
	(assert (= (quartiles [13])
	[13 13 13 13 13]))
	(assert (= (quartiles (range 100001))
	[0 25000 50000 75000 100000]))
	(assert (nil? (quartiles [])))
	true)