Bulletproof statistical functions

dev.metalisp.survey.asd

@@ -2,7 +2,7 @@
 
 (defsystem "dev.metalisp.survey"
   :description "A simple survey"
-  :version "0.4.6"
+  :version "0.4.7"
   :author "Marcus Kammer <marcus.kammer@metalisp.dev>"
   :source-control "git@git.sr.ht:~marcuskammer/dev.metalisp.survey"
   :licence "MIT"

src/stats.lisp

@@ -10,6 +10,9 @@
 
 (in-package #:ml-survey/stats)
 
+(define-condition stats-error (error)
+  ((message :initarg :message :reader error-message)))
+
 (defun preprocess-and-transpose (data)
   (apply #'mapcar #'list (mapcar #'cdr data)))
 
@@ -21,19 +24,34 @@
         (reduce #'max numbers)))
 
 (defun standard-deviation (numbers)
-  (let* ((avg (mean numbers))
-         (variance (/ (reduce #'+ (mapcar (lambda (x) (expt (- x avg) 2)) numbers))
-                      (1- (length numbers)))))
-    (sqrt variance)))
+  (handler-case
+      (let ((len (length numbers)))
+        (if (< len 2)
+            (error 'stats-error :message "Need at least two numbers for standard deviation")
+            (let ((avg (mean numbers)))
+              (sqrt (/ (reduce #'+ (mapcar (lambda (x) (expt (- x avg) 2)) numbers))
+                       (1- len))))))
+    (type-error ()
+      (error 'stats-error :message "Invalid input for standard deviation calculation"))))
 
 (defun mean (numbers)
+  (handler-case
+      (if (null numbers)
+          (error 'stats-error :message "Cannot calculate mean of an empty list")
           (/ (reduce #'+ numbers) (length numbers)))
+    (division-by-zero ()
+      (error 'stats-error :message "Division by zero in mean calculation"))))
 
 (defun median (numbers)
+  (handler-case
       (let ((sorted (sort (copy-seq numbers) #'<))
-        (count (length numbers)))
-    (if (oddp count)
-        (nth (floor count 2) sorted)
-        (/ (+ (nth (1- (/ count 2)) sorted)
-              (nth (/ count 2) sorted))
-           2.0))))
+            (len (length numbers)))
+        (if (zerop len)
+            (error 'stats-error :message "Cannot calculate median of an empty list")
+            (if (oddp len)
+                (nth (floor len 2) sorted)
+                (/ (+ (nth (1- (/ len 2)) sorted)
+                      (nth (/ len 2) sorted))
+                   2))))
+    (type-error ()
+      (error 'stats-error :message "Invalid input for median calculation"))))