PCA_KMeans_demo.py

import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D

from sklearn.decomposition import PCA
from sklearn.cluster import KMeans
from sklearn import datasets

import brica1

# PCA Component Definition
class PCAComponent(brica1.Component):
    def __init__(self,n_in,dimention=3):
        super(PCAComponent, self).__init__()
        self.pca = PCA(n_components=dimention)
        self.make_in_port("in0",n_in)
        self.make_out_port("out0",dimention)
    
    def fire(self):
        x = self.inputs["in0"]
        z = self.pca.transform(x)
        self.results["out0"] = z
        
    def fit(self,dataset):
        return self.pca.fit_transform(dataset)

# KMeans Component Definition
class KMeansComponent(brica1.Component):
    def __init__(self, n_in,n_clusters=3 ):
        super(KMeansComponent, self).__init__()
        self.kmeans = KMeans(n_clusters = n_clusters )
        self.make_in_port("in0", n_in)
        self.make_out_port("out0", 1)
 
    def fire(self):
        x = self.inputs["in0"]
        z = self.kmeans.predict(x)
        self.results["out0"] = z
 
    def fit(self, dataset):
        self.kmeans.fit(dataset)


# Load iris dataset
iris = datasets.load_iris()
X = iris.data
y = iris.target

# Setup data feeder component
feeder = brica1.ConstantComponent()
feeder.make_out_port("out0", len(X[0]))

# Setup PCA component
dim = 3
pca = PCAComponent(n_in=len(X[0]), dimention = dim)
Z = pca.fit(X)

# Setup KMeans Component
kmeans = KMeansComponent(n_in= dim)
kmeans.fit(Z)

# Connect the components
brica1.connect((feeder,"out0"),(pca,"in0"))
brica1.connect((pca,"out0"),(kmeans,"in0"))

# Add components to the module
mod = brica1.Module()
mod.add_component("feeder",feeder)
mod.add_component("pca",pca)
mod.add_component("kmeans",kmeans)

# Setup scheduler and agent
s = brica1.VirtualTimeSyncScheduler()
a = brica1.Agent(s)
a.add_submodule("mod",mod)

# Test the pca and kmeans clustering
kmeans_results=[]
for i in xrange(len(X)):
    feeder.set_state("out0",X[i])
    
    a.step()
    a.step()
    
    kmeans_results.append(kmeans.get_out_port("out0").buffer[0])

hoge =[]
for e in kmeans_results:
    if e ==1:
        hoge.append(0)
    elif e ==2:
        hoge.append(1)
    elif e ==0:
        hoge.append(2)

for i in xrange(len(X)):
    print "Label: {}\tKMeans_Result: {}\tTrue or Faluse: {}".format(y[i], hoge[i], y[i]==hoge[i])