335K - views

Journal of Mathematical Psychology Tutorial Tutorial on maximum likelihood estimation In Jae Myung Department of Psychology Ohio State University Neil Avenue Mall Columbus OH USA Received Novem

Theintendedaudienceofthistutorialare researchers who practice mathematical modeling of cognition but are unfamiliar with the estimation method 1nlike least2squares estimation which is primarily a descriptive tool M is a preferred method of parameter

Embed :
Pdf Download Link

Download Pdf - The PPT/PDF document "Journal of Mathematical Psychology Tu..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.

Journal of Mathematical Psychology Tutorial Tutorial on maximum likelihood estimation In Jae Myung Department of Psychology Ohio State University Neil Avenue Mall Columbus OH USA Received Novem






Presentation on theme: "Journal of Mathematical Psychology Tutorial Tutorial on maximum likelihood estimation In Jae Myung Department of Psychology Ohio State University Neil Avenue Mall Columbus OH USA Received Novem"— Presentation transcript:

          !    " #DepartmentofPsychology,OhioStateUniversity,1885NeilAvenueMall,Columbus,OH43210-1222,USA$  % !&% ' ( % !)* ' " ++ ,"+%!    +      !     -./  ! !!      �#?B)22C)/ /D/ !/2)EEF2   . % 4  145/5   % !/�!/)E42)A2 "   ++ ," !     !     !    ! 2/ ++ 0    0  + %% ! +0 !+'  +   G //JournalofExperimentalPsychology' ! ! / ++  ! !  %  ++   !    % ' !   +  -4.    ! !  ,  ' +!   +         , + 2  ! / ++  ++  +%! ! + !  ! !0   + /?2! + ,         +,  ! !  !    //,@ :H  ,,= +/(= :@  ,,= +/(H 7 :4 % ,,= +/)(4+,,= +//   2.1.Probabilitydensityfunction?     !+ , ! %  !  +   0++ / !  !   ++        %    !  + /"   ,  ++ !  6 !' +!+'' ! ' /5 !0  +'' ! ' 3 %  ! I+ /5 +   % ,!   +'' ! '     !/?2 , ! ! 6 !  +'' ! ' !  !' ! I+ /!  probabilitydensityfunctionH?  + 6  +'' ' %! %  +    ++ 0 0 + %   //! 0 '+ %      /// + %  ! 6 !  2! + + /"!%!'2 % ,I,    ! + !    , !  +'' , H? ! %  + ' +  !  + H?!%!' % ,  ! H?,!   +   0  ' % ! + ,  ,4++  !  +     '    3  @   //   !  +''   , +   !' + // H?  % ' yyyAyyy ?//@ +'' ! '  + !+'' +  +!'  /I.J.Myung/JournalofMathematicalPsychology47(2003)90100 0   0 ' ! ' 0 + &   ' !  !+ /  +  H? 0 ++ ?//" + %   !  0H?'  ! 0   +  0 '  + ?// 0    +  H? ' ! ' ' %  0   + 6  +'' ! % %     H?   !'% +     ! 6  ! /2.2.Likelihoodfunction7%  + % ,  +!H?0 0   !   +''  ! /" + %  + , H?0      C/%//)/"  , 0 % ,0 %  !' % ! ! /5!,0   !0 %  +' �7%  ' % !! ! !    ,6!  H?,  +'' !     ! + ' ,       % +! ! ! /% %  +' ,0 ! 6 likelihoodfunction' %    ! %  ! + %   +      ! + %  ' % !! !  ?  2+ '   + .3/,   ! !% '   +    !  0?/    +  !   ' 0  H?!   ! 5  !?/!, 0  ! 6 !!    ,!     !   +'    /4+ 6, H??/  ! % +  + % ,! 6 ! datascale/*   !,   !   + % +  ' % !! ,! 6 ! parameterscale/"  ,?/  +'' + ! % 6 !+ ,0  ?/    !JJ +''2 II+ + % 6 !!  /&    !  6 % ?//   ! % ' % !! ! +   2+  ! ! ' !  /I.J.Myung/JournalofMathematicalPsychology47(2003)90100 '    + ' ! 0   ! ' 0/" !   0+ ,   ! 0'   '% + + /"  , ! 0 + ,   !     + 2!   JJ II '% 2! + + + !' + %     * !  % '   !!   !  ! %  ! !   !, +          ' ++ ,  , +'' ! '   !   ! /7%   !   + % ! !   +'' ! ' ?/,0     !6! + %   2+!  !  !+'' ! ' / ++ maximumlikelihoodestimation -.,! % + !'$/5/?  ,    !  !+'' ! '      ' % !! JJ   ,II0       %  + %        !    + %  ,0   '    2! + + , ! MLEestimate,!!  !'?  + ,?/, -.    0    2 !%  +'' ! '  +!   -.     0 '  + ?/5!  -.++ ,  ++        %    ! ' % !!  ,     !   2  !  +'' ! '     ' % !!    /3.1.Likelihoodequation -.     !  ' 3 /"   ,0  0 0  + -.    0    ! 3 /? +  % 2  , -.    '  !'   2  ! , '  0 ,    !      -.    '  !'     /5   2  ! ,!   ' ,  ,    0+ !    3  0 likelihoodequation  '  ! 6        !   '   +    6 ! % % %   + /   ! 3  +     2!     -.    /5!! !   '  6 !       !    , 6 ! % %   %  /'    ,  +  2  !  !' %    +   +  , %    ' ! '   !'   !! % %  2  !! 00     %    -.+ ! , !  + % 2+ '   + % 6 !% ? ,'       ! .3/),0 '  2  ! &  , 6 ! % %  2  ! ! dLwnydw w w @ 3  3  ' !  ! -.    '  !      +      ,    ,  !! % %  2  ! !! % ! dLwnydw w 0    % ,!  !/"+  , 0 % ,  +' '     -.    , + 0  ! %%  + ! H?  2 /"   , -.     '   2 +    / '! 2 +    3 6!+  +     2  !/ ! ' =!  9   ! 6 !       %  !  3   !   negativedenite  , 2 '  !%  !   + I.J.Myung/JournalofMathematicalPsychology47(2003)90100       '2   2! 2+ +      %   2 0  + + ,0  '    '   ' +   / JJ   II  + !' ! %       %  +/4+ 6,    ,'       + %  , 0 + % '  !'!!     + %+  0   0+     !  +% !+   /H   +    !  0 +!   ! !/   % + , 0'  0?/,    +  G! !  % %  !/ /,+ @?/ +   + +++ + ! 6 ! /. +   ++ !    '   0 !       + % ' 0  0 %   /3.2.Localmaxima" 0    +    !       + %   3    2  !0' !/?!+   +      + 0  +       +% +  +   ++ !'  /" + %       ! ' /H + !+     + % ,  !+    +! '2+   + % /  ! localmaxima+' /5  + ,?/      + %  + 0 !  +  + @ ! globalmaximum,   + %  + 0 ! + 5,0  '2+   /4 ,   + %  0 !  '2+    + =/1  ,           +' /" !,%   23  % ' ! % + ! + %! +' ,        %  /?  + ,   !     % %   +    + ! !        0    '  ! +  !/;   ++ , ! 0  6!   '   ' !/ ?//5  +  2  ! 6   2+  ! / @ '   0  + 5!= 0  /   0! +   % +   + / 5   +    0  ! K +  ,7  ,:L  ,A%    +' ,    ,    '  ' + +         6! I.J.Myung/JournalofMathematicalPsychology47(2003)90100 3.3.Relationtoleast-squaresestimation$    -.0  + %   mostlikely  % +! ! ! /"-4.,   !,0  + %   +%! mostaccurate! +  ! ,  !   0  ! 6  ! !  3 2 /? ,-4.,sumofsquareserror44.' 0 ' % !+ !  &#x !00; !   ! I+ !  ' % /&    + %  5 -.,6! + %     44.   3  2 +    /   -4.'G     +' , + 0  ! 2 0  +   + /   ' 0  0 !     % 2 % 3  /"  ,-4.     ! !  -.    , + !      ! ' ! ++   ! +  /5 +     +'%  !   '  !  ! + !+0  !     + ! ! /;  , -. !' +   ! -4.,  +'' !    0!6 '   + '  ,  , ! !    $ ,A    , 0 % ,0  0 !   / 0 ' %  ! + !    !  ! ' !0   % /"  ,    2  ! 3%     44.,!   ,  + %  '  !!    -.-4./ "   ,"+   ++   +      !    /  !,"   ! %           + //$':;  ,); ,A; !:.''  ,5  2!     % ' ++ !!  ! + ,"+%!   +  -. 0 ,+0 ! +  /- ' + %  ,  ,! ! I+ !  +'2'       0 !  ! 6 !+0  ! &#x -5;7.2; +   ! &#x -5;7.2;4++  !  ' % 0   +   '2 % !++      ;     !   %'   ! /; ! '6   ' % !! !   ! 6 /5++  -. 3 + 6  H? ! undereachmodel/! ,6 0    ' % !++ '  !'!%!  '    + '   ' ! + !  ;    ' ! ' !0 +''   H? +0  ! ! +   !  '  !+0 &#x -5;7.2; +  � 0     0+  ' !  ! H? '% 3 /? , +'' + ' +'' ! ' / /.3/'  !  !/   , H?  ! .3/'  !' + + +'' + .3/0 !  3 ,.3//4 !,    ! '6 !  ,5 ,    !++' !     +     /"+ , H?'  !' + +&0,   I    ! + !    , !  !2  !  +0  ! % ' ?  ! , H? +  !6    !   /I.J.Myung/JournalofMathematicalPsychology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yung/JournalofMathematicalPsychology47(2003)90100  ++ !  ' ! ' !,  ! ' !/? ,   %  +2  3 ! 3%  ' 0  -.!-4.!  !' ! 0  % ,! + !+++   4.1.MLEinterpretation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yung/JournalofMathematicalPsychology47(2003)90100     6      #        # 9     "   # 6*:   *: " #   "*:      "*.   6,!;' (!;! �    � �?=     "        #    6  - 6    "    - "       $     $  - 76*6- -8@$      - 7*-"*"-8@$              �=  #    6     "$ 6        .   #             #       )=  #   "       "$"        .   #             #    ! "$#     /                  !          9    $      %&  '(% (%&  (% (% )(*#+% , (*#+%&  #(% (% "! (-..% / (% ($                   *1+2*-*3         I.J.Myung/JournalofMathematicalPsychology47(2003)90100 %! 5  ,9//"   !       !++ /�" ,@/&/,:= ,Secondinternationalsymposiumoninformationtheory++/)A/�@!+  5 ! K!/@  ! ,;/9/,:=0 ,=/4//   +   !    /JournalofMathematical,CCC/@ ,//,:H  ,K/5//Mathematicalstatistics* !,&#x12.; =59! 2!,"/= ,7/,:@  ,$/-//Statisticalinference! !//67% ,&#x12.; =5H' /H 7 , /9/,:4 % , ///Probabilityandstatistics! !//@ ,&#x12.;  55!!2;  /K +  ,4/,7  ,=/H/,:L  , //A/*+   '  ! /,))A/- '  ,K//"  2   + ! !  /PsychologicalReview,)/-  ,9/,:N ,;/A)/Modelselection/& 0O ,&#x12.; &O ! /,@/@/)/   !%! /JournalofExperimentalPsychology,)A)C/ ,"//,? , /,:@0 , /;//4+   !    /JournalofMathematicalPsychology , /5/, ,"//,:N ,4//0! !     +   !  /Psychological,/$ ,$/A/5    %/Psychological,CA/$',H/=/,9 ,4/,:; 5// +      /JournalofExperimentalPsychology:Learning,Memory,andCognition,))/                     6       "*9  % 6   6 6 6     6  6*:/ *: //   *:         "*:      "*.   6"-9    6 6 6     "    - 6  - "       $  - 76*6- -8@$       - 7*-"*"-8@$              �= /  6       6    $ 6      6      9  6     )= /  6      "      $"        6      9  6  I.J.Myung/JournalofMathematicalPsychology47(2003)90100 $',H/=/,:; 5/./)/* ! !  &#x ,-;ٕ.; 53   % ! +   /Psychological,)/4 07/A/.    !  ! /Annalsof,))/4+,5//Probabilitytheoryandstatisticalinference/= 2'! ,&#x,-37;䀀1K= '! 1%   /1 , /,: = !,/-//   +  +   (   , +    ! /Psychological,CCC/LN! ,//90 6  +  ! ' /PsychonomicBulletin&Review,)C/; ,/H/A/*     &#x,-37;䀀 =  $'!; &#x -2;p.8;)53   % ! +   /PsychologicalReview,A)/; !,//,:.''  ,./@//*   /PsychologicalScience,C/I.J.Myung/JournalofMathematicalPsychology47(2003)90100