trips.m

交流 模糊控制 交流 模糊控制

%% Model Identification
% This is a demonstration of subtractive clustering and how it can be used
% with multi-dimensional data.
%
% Copyright 1994-2002 The MathWorks, Inc. 
% $Revision: 1.12 $

%%
% This is a plot of the input data for a model identification
% problem. We are interested in estimating the number
% of auto trips generated from an area based on the area's
% demographics. Five factors were considered: population,
% number of houses, vehicle ownership, income, and employment.

tripdata
plot(datin)

%%
% Using the GENFIS2 function (which is based on the
% subtractive clustering algorithm in the SUBCLUST function),
% we generate a fuzzy inference system that calculates the
% output based on the five inputs.

a=genfis2(datin,datout,0.45);
plotfis(a);

%%
% The upper plot displays 75 data points for the five input
% variables. The lower plot displays the corresponding outputs
% and the outputs predicted by the fuzzy model.

subplot(1,1,1)
fuzout=evalfis(datin,a);
subplot(2,1,1)
plot(datin)
subplot(2,1,2)
plot([datout fuzout])

%%
% Here is a plot of the actual output values (X axis) versus the
% predicted output values (Y axis). If the prediction were a
% perfect one, the data points would lie right along the diagonal
% line X = Y.

subplot(1,1,1)
plot(datout,fuzout,'bx',[0 10],[0 10],'r:')
xlabel('Actual Value')
ylabel('Predicted Value')
axis square

%%
% We set aside 25 of the original 100 data points as checking
% data. Since we did not use this data to create our model, it
% is a useful measure of how good our model is.

chkfuzout=evalfis(chkdatin,a);
plot(chkdatout,chkfuzout,'bx',[0 10],[0 10],'r:')
axis square
xlabel('Actual Value')
ylabel('Predicted Value')

%%
% Clustering can be a very effective technique for dealing with
% large sets of data: the principal idea is to distill natural
% groupings of data from a large data set thereby allowing
% concise representation of a model's behavior. This demo has
% shown how accurate predictions can be made despite the
% multi-dimensional nature of the problem. With the results of this
% clustering experiment in hand, we could now potentially go on
% to use other techniques, such as ANFIS.