flucoma-sc/release-packaging/HelpSource/Classes/FluidKNNClassifier.schelp

TITLE:: FluidKNNClassifier
summary:: Classify data with K Nearest Neighbours
categories:: Classification, KNN
related:: Classes/FluidKNNRegressor, Classes/FluidDataSet, Classes/FluidLabelSet

DESCRIPTION::
A nearest-neighbour classifier using link::Classes/FluidKDTree:: . Each point is assigned the class that is most common among its nearest neighbours.
https://scikit-learn.org/stable/modules/neighbors.html#classification

CLASSMETHODS::

METHOD:: new
Create a new KNNClassifier
ARGUMENT:: server
The server to make the model on
ARGUMENT:: numNeighbours
the number of neighours to consider
ARGUMENT:: weight
true / false: whether the neighbours should be weighted by distance

INSTANCEMETHODS::

METHOD:: fit
Fit the model to a source link::Classes/FluidDataSet:: and a target link::Classes/FluidLabelSet::. These need to be the same size
ARGUMENT:: dataSet
Source data
ARGUMENT:: labelSet
Labels for the source data
ARGUMENT:: action
Run when done

METHOD:: predict
Given a fitted model, predict labels for a link::Classes/FluidDataSet:: and write these to a link::Classes/FluidLabelSet::
ARGUMENT:: dataSet
data to predict labels for
ARGUMENT:: labelSet
place to write labels
ARGUMENT:: action
Run when done

METHOD:: predictPoint
Given a fitted model, predict labels for a data point in a link::Classes/Buffer:: and return these to the caller
ARGUMENT:: buffer
A data point
ARGUMENT:: action
Run when done, passes predicted label as argument

EXAMPLES::

code::


// Make:
// - A KNN Classifier
// - A DataSet of example points, and a label set of corresponding labels
// - A DataSet of test data and a LabelSet for predicted labels

(
~classifier = FluidKNNClassifier(s);
~source= FluidDataSet(s,\knnclassify_help_examples);
~labels = FluidLabelSet(s,\knnclassify_help_labels);
~test = FluidDataSet(s,\knnclassify_help_test);
~mapping = FluidLabelSet(s,\knnclassify_help_mapping);
)

//Make some clumped 2D points and place into a DataSet
(
~examplepoints = [[0.5,0.5],[-0.5,0.5],[0.5,-0.5],[-0.5,-0.5]];
~examplelabels = [\red,\orange,\green,\blue];
d = Dictionary.new;
d.add(\cols -> 2);
d.add(\data -> Dictionary.newFrom(~examplepoints.collect{|x, i|[i.asString, x]}.flatten));
~source.load(d);
~examplelabels.collect{|x,i| ~labels.addLabel(i, x);};
)

//Make some random, but clustered test points
(
~testpoints = (4.collect{
               64.collect{(1.sum3rand) + [1,-1].choose}.clump(2)
           }).flatten(1) * 0.5;
d = Dictionary.with(
        *[\cols -> 2,\data -> Dictionary.newFrom(
            ~testpoints.collect{|x, i| [i, x]}.flatten)]);
 ~test.load(d);

)


//Fit the classifier to the example DataSet and labels, and then run prediction on the test data into our mapping label set
(
~classifier.fit(~source,~labels);
~classifier.predict(~test, ~mapping, 1);
)

//Return labels of clustered points
(
~assignments = Array.new(~testpoints.size);
fork{
    ~testpoints.do{|x,i|
        ~mapping.getLabel(i, action:{|l|
            ~assignments.add(l);
        });
        s.sync;
        if(i==(~testpoints.size - 1)){"Got assignments".postln;}
    };
    ~assignments.postln;
}
)

//Visualise: we're hoping to see colours neatly mapped to quandrants...
(
c = IdentityDictionary();

c.add(\red->Color.red);
c.add(\blue->Color.blue);
c.add(\green->Color.green);
c.add(\orange-> Color.new255(255, 127, 0));

e = 200 * ((~examplepoints + 1) * 0.5).flatten(1).unlace;
d = ((~testpoints + 1) * 0.5).flatten(1).unlace;
// d = [20.collect{1.0.rand}, 20.collect{1.0.rand}];
w = Window("scatter", Rect(128, 64, 200, 200));
~colours = [Color.blue,Color.red,Color.green,Color.magenta];
w.drawFunc = {
    Pen.use {
        e[0].size.do{|i|
            var r = Rect(e[0][i],e[1][i],10,10);
            Pen.fillColor = c[~examplelabels[i]];
            Pen.fillOval(r);
        };
        d[0].size.do{|i|
            var x = (d[0][i]*200);
            var y = (d[1][i]*200);
            var r = Rect(x,y,5,5);
            Pen.fillColor = c[~assignments[i].asSymbol].alpha_(0.3);
            Pen.fillOval(r);
        }
    }
};
w.refresh;
w.front;
)
::

subsection::Server Side Queries
This is the equivalent of predictPoint, but wholly on the server
FluidKNNClassifier is accessed via its own synth, so we need to use
a bus to communicate with it. The inBus receives a trigger to query, using data
from inBuffer; a trigger is then send to outBus with the prediction in outBuffer
code::
(
~ib = Bus.control(s); // input bus
~ob = Bus.control(s); //output bus
~inpPoint = Buffer.alloc(s,2);
~outPoint = Buffer.alloc(s,1);
)

//We make two Synths. One, before FluidKMeans, generates a random point and sends
//a trigger to query. The second, after FluidKMeans, gives us the predicted cluster //triggering upadtes from the outBus
(
//Set properties on FluidKNNClassifier:
~classifier.inBus_(~ib).outBus_(~ob).inBuffer_(~inpPoint).outBuffer_(~outPoint);
//pitching
{
	var trig = Impulse.kr(5);
	var point = WhiteNoise.kr(1.dup);
	Poll.kr(trig, point, [\pointX,\pointY]);
	point.collect{ |p,i| BufWr.kr([p],~inpPoint,i)};
	Out.kr(~ib.index,[trig]);
}.play(~classifier.synth,addAction:\addBefore);
//catching
{
	Poll.kr(In.kr(~ob),Latch.kr(BufRd.kr(1,~outPoint,0,interpolation:0),In.kr(~ob)),\cluster);
}.play(~classifier.synth,addAction:\addAfter);
)

// to sonify the output, here are random values alternating quadrant.
(
//Set properties on FluidKNNClassifier:
~classifier.inBus_(~ib).outBus_(~ob).inBuffer_(~inpPoint).outBuffer_(~outPoint);
//pitching
{
	var trig = Impulse.kr(MouseX.kr(0,1).exprange(0.5,ControlRate.ir /2).poll);
	var step = Stepper.kr(trig,max:3);
	var point = TRand.kr(-0.1, [0.1, 0.1], trig) + [step.mod(2).linlin(0,1,-0.6,0.6),step.div(2).linlin(0,1,-0.6,0.6)] ;
	point.collect{|p,i| BufWr.kr([p],~inpPoint,i)};
	Out.kr(~ib.index,[trig]);
	T2A.ar(trig)*0.1;
}.play(~classifier.synth,addAction:\addBefore);
//catching
{
	SinOsc.ar((Latch.kr(BufRd.kr(1,~outPoint,0,interpolation:0),In.kr(~ob)) + 69).midicps,mul: 0.1);
}.play(~classifier.synth,addAction:\addAfter);
)
::