;;;; -*- mode: common-lisp; coding: utf-8; -*-

(defpackage ml-survey/stats
  (:use #:cl)
  (:export #:preprocess-and-transpose
           #:calculate-statistics
           #:stdev
           #:average
           #:median
           #:geomean))

(in-package #:ml-survey/stats)

(defun preprocess-and-transpose (data)
  (apply #'mapcar #'list (mapcar #'cdr data)))

(defun calculate-statistics (numbers)
  (list :median (median numbers)
        :average (average numbers)
        :geomean (geomean numbers)
        :stdev (stdev numbers)
        :min (reduce #'min numbers)
        :max (reduce #'max numbers)))

(defun average (numbers)
  (if (null numbers)
      0
      (* 1.0 (/ (reduce #'+ numbers)
                (length numbers)))))

(defun geomean (numbers)
  (if (null numbers)
      0
      (expt (reduce #'* numbers)
            (/ 1.0 (length numbers)))))

(defun median-odd (numbers)
  (nth (floor (length numbers) 2) numbers))

(defun median-even (numbers)
  (let* ((mid (floor (length numbers) 2))
         (pair (list (nth (1- mid) numbers)
                     (nth mid numbers))))
    (average pair)))

(defun median (numbers)
  (let ((sorted (sort (copy-list numbers) #'<)))
    (if (oddp (length sorted))
        (median-odd sorted)
        (median-even sorted))))

(defun range (numbers)
  (- (reduce #'max numbers)
     (reduce #'min numbers)))

(defun variance (numbers)
  (if (< (length numbers) 2)
      nil  ; or we could return nil to indicate undefined variance
      (let* ((n (length numbers))
             (average-value (average numbers))
             (squared-diff (mapcar (lambda (x) (expt (- x average-value) 2)) numbers)))
        (/ (reduce #'+ squared-diff) (1- n)))))

(defun stdev (numbers)
  (sqrt (variance numbers)))