/*----------------------------------------------------------*/
/* Weather data set with numeric attributes                 */
/* T. Mitchell, Machine Learning, McGraw-Hill, 1997         */
/*----------------------------------------------------------*/

example(1, no, [outlook=sunny,   temp=85,humidity=85,windy=f]).
example(2, no, [outlook=sunny,   temp=80,humidity=90,windy=t]).
example(3, yes,[outlook=overcast,temp=83,humidity=86,windy=f]).
example(4, yes,[outlook=rainy,   temp=70,humidity=96,windy=f]).
example(5, yes,[outlook=rainy,   temp=68,humidity=80,windy=f]).
example(6, no, [outlook=rainy,   temp=65,humidity=70,windy=t]).
example(7, yes,[outlook=overcast,temp=64,humidity=65,windy=t]).
example(8, no, [outlook=sunny,   temp=72,humidity=95,windy=f]).
example(9, yes,[outlook=sunny,   temp=69,humidity=70,windy=f]).
example(10,yes,[outlook=rainy,   temp=75,humidity=80,windy=f]).
example(11,yes,[outlook=sunny,   temp=75,humidity=70,windy=t]).
example(12,yes,[outlook=overcast,temp=72,humidity=90,windy=t]).
example(13,yes,[outlook=overcast,temp=81,humidity=75,windy=f]).
example(14,no, [outlook=rainy,   temp=71,humidity=91,windy=t]).

/* Extracts from the above data provided for convinient access */

/* Nominal-only data */

nominal(1, no,   [outlook=sunny, windy=f]).
nominal(2, no,   [outlook=sunny, windy=t]).
nominal(3, yes,  [outlook=overcast, windy=f]).
nominal(4, yes,  [outlook=rainy, windy=f]).
nominal(5, yes,  [outlook=rainy, windy=f]).
nominal(6, no,   [outlook=rainy, windy=t]).
nominal(7, yes,  [outlook=overcast, windy=t]).
nominal(8, no,   [outlook=sunny, windy=f]).
nominal(9, yes,  [outlook=sunny, windy=f]).
nominal(10, yes, [outlook=rainy, windy=f]).
nominal(11, yes, [outlook=sunny, windy=t]).
nominal(12, yes, [outlook=overcast, windy=t]).
nominal(13, yes, [outlook=overcast, windy=f]).
nominal(14, no, [outlook=rainy, windy=t]).

/*-----------------------------------------------------------*/
/* Numeric data extracted from the original set of examples  */
/*-----------------------------------------------------------*/

/* temp_sample(<List of values form class yes>, <List of values from class no>) */

temp_sample([83, 70, 68, 64, 69, 75, 75, 72, 81], [85, 80, 65, 72, 71]).

/* humidity_sample(<List of values form class yes>, <List of values from class no>) */

humidity_sample([86, 96, 80, 65, 70, 80, 70, 90, 75], [85, 90, 70, 95, 91]).

temp(1, no, 85).
temp(2, no, 80).
temp(3, yes, 83).
temp(4, yes, 70).
temp(5, yes, 68).
temp(6, no, 65).
temp(7, yes, 64).
temp(8, no, 72).
temp(9, yes, 69).
temp(10, yes, 75).
temp(11, yes, 75).
temp(12, yes, 72).
temp(13, yes, 81).
temp(14, no, 71).

humidity(1, no, 85).
humidity(2, no, 90).
humidity(3, yes, 86).
humidity(4, yes, 96).
humidity(5, yes, 80).
humidity(6, no, 70).
humidity(7, yes, 65).
humidity(8, no, 95).
humidity(9, yes, 70).
humidity(10, yes, 80).
humidity(11, yes, 70).
humidity(12, yes, 90).
humidity(13, yes, 75).
humidity(14, no, 91).

%----------------------------------------------
%  Examples of using the data set with stats.pl
%----------------------------------------------

%--------------------------------------------------------
%  Compute the probability density of 67 for both classes
%   P(67|Yes) = 0.0402995
%   P(67|No) = 0.0317938
%--------------------------------------------------------
/*
?- temp_sample(Yes,No),p(67,Yes,PYes),p(67,No,PNo).

Yes = [83, 70, 68, 64, 69, 75, 75, 72, 81]
No = [85, 80, 65, 72, 71]
PYes = 0.0402995
PNo = 0.0317938 
*/

%-----------------------------------------------------
%  Compute the probability density of values from temp
%-----------------------------------------------------
/*
?- temp_sample(Yes,No),temp(N,Class,X),p(X,Yes,PYes),p(X,No,PNo).

Yes = [83, 70, 68, 64, 69, 75, 75, 72, 81]
No = [85, 80, 65, 72, 71]
N = 1
Class = no
X = 85
PYes = 0.00973071
PNo = 0.0212165 ;

Yes = [83, 70, 68, 64, 69, 75, 75, 72, 81]
No = [85, 80, 65, 72, 71]
N = 2
Class = no
X = 80
PYes = 0.0339635
PNo = 0.0399971 ;

...

*/

% Predict and compute probabilities with nominal and mixed attributes

% Note that bayes.pl works only with discrete probabilities (nominal attributes)

/*
?- nominal(N,C,E),bayes(E,P),write(N-C-P),nl,fail.
1-no-yes
2-no-no
3-yes-yes
4-yes-yes
5-yes-yes
6-no-no
7-yes-yes
8-no-yes
9-yes-yes
10-yes-yes
11-yes-no
12-yes-yes
13-yes-yes
14-no-no


?- nominal(N,C,E),probs(E,P),write(N-C-P),nl,fail.  
1-no-[no/0.0857143, yes/0.0952381]
2-no-[no/0.128571, yes/0.047619]
3-yes-[no/0, yes/0.190476]
4-yes-[no/0.0571429, yes/0.142857]
5-yes-[no/0.0571429, yes/0.142857]
6-no-[no/0.0857143, yes/0.0714286]
7-yes-[no/0, yes/0.0952381]
8-no-[no/0.0857143, yes/0.0952381]
9-yes-[no/0.0857143, yes/0.0952381]
10-yes-[no/0.0571429, yes/0.142857]
11-yes-[no/0.128571, yes/0.047619]
12-yes-[no/0, yes/0.0952381]
13-yes-[no/0, yes/0.190476]
14-no-[no/0.0857143, yes/0.0714286]
*/


% Computing probabilities with all 4 attributes (nominal and numeric)
/*
?- temp_sample(TY,TN),humidity_sample(HY,HN),temp(N,C,T),humidity(N,C,H),
   p(T,TY,PTY),p(T,TN,PTN),p(H,HY,PHY),p(H,HN,PHN), % Compute densities
   nominal(N,C,E),probs(E,[no/PN,yes/PY]),          % Compute discrete probs
   LY is PTY*PHY*PY, LN is PTN*PHN*PN,              % Compute likelihoods
   PPY is LY/(LY+LN), PPN is LN/(LY+LN),            % Normalize
   write(N-PPY-PPN),nl,fail.                        % Print

1-0.292919-0.707081
2-0.154835-0.845165
3-1-0
4-0.576106-0.423894
5-0.791622-0.208378
6-0.711311-0.288689
7-1-0
8-0.388032-0.611968
9-0.791346-0.208654
10-0.779496-0.220504
11-0.535425-0.464575
12-1-0
13-1-0
14-0.380395-0.619605
*/