examples/50-TTestExample.lisp

   1 ;;; -*- mode: lisp -*-
   2
   3 ;;; Time-stamp: <2012-10-12 10:44:00 tony>
   4 ;;; Creation:   <2009-04-19 09:41:09 tony>
   5 ;;; File:       t-test-example.lisp
   6 ;;; Author:     AJ Rossini <blindglobe@gmail.com>
   7 ;;; Copyright:  (c)2009--, AJ Rossini.  MIT
   8 ;;; Purpose:    Example of basic exploratory data analysis in CLS.
   9
  10 ;;; What is this talk of 'release'? Klingons do not make software
  11 ;;; 'releases'.  Our software 'escapes', leaving a bloody trail of
  12 ;;; designers and quality assurance people in its wake.
  13
  14
  15 ;;; If we can't do a t-test, we can't do anything
  16
  17 ;;;; BROKEN!!!
  18
  19 (in-package :cls-user)
  20
  21 (defparameter *my-df*
  22   (make-dataframe
  23    (rsm.string::filestream->string-table
  24     "/path/to/file.dsv"))
  25   "Initial read-in of data.")
  26
  27
  28 ;;;; possible APIs...
  29
  30 #|
  31
  32 summarize what how send-to
  33
  34 what: df, stat-proc
  35 how:  numerically, visually, all, default
  36 send-to: output, file, visual, all, default
  37
  38
  39 compare in-model parameters framework philosophy assumptions
  40
  41 in-model: regression two-group
  42 parameters: means, "explicit choices", variances, means-and-variances, rates, proportions, hazards
  43 framework: fisherian, neyman-pearson, empirical-bayes
  44 philosophy: frequentist, bayesian, fiducial, decision-theoretic,
  45 assumptions: (list 'apply 'ignore 'explore) ; any or all of list entries
  46
  47
  48 FIXME: where would empirical bayes go?  part is a model, part is a philosophy.  framework?
  49 FIXME: decision-theoretic adds risk to the decision, making
  50
  51 the in-model objects need to be able to reflect their own assumptions and preferences,
  52
  53 |#
  54
  55
  56 ;;; The traditional intro-stat variant
  57
  58 (summarize
  59  (compare :parameter 'mean
  60           :comparison 'two-groups
  61           :group1 datapackage:variable1
  62           :group2 datapackage:variable2
  63           :assumptions (list 'apply 'ignore 'explore) ; any or all of list entries
  64           )
  65  :resulttype (list 'numerical 'visual)
  66  )  ; any or all of list entries
  67
  68 ;;; the "know-your-dataset" variant
  69
  70 (summarize
  71  (compare :parameter 'mean
  72           :resulttype (list 'numerical 'visual)
  73           :comparison 'two-groups
  74           :variable datapackage:comparing-variable ; or response-variable
  75           :group-by datapackage:grouping-variable))
  76