def image_generation(us_mlp, s, w, h, file_name):
in_vect= [0.0, 0.0, 0.0, 0.0]
out_vect=[0.0,0.0,0.0]
in_vect[2]= s[0]*(max_0-min_0)+min_0
in_vect[3]= s[1]*(max_1-min_1)+min_1
svg = MYSVGWriter(640, 480, 0, 0, w, h)
for y in range(h):
in_vect[1] = 1.0* y/h
for x in range(w):
in_vect[0] = 1.0*x/w
out_vect = us_mlp.predict(in_vect)
svg.rect(x, h - 1 - y, 1, 1, rgbToUint(out_vect[0] * 255, out_vect[1] * 255, out_vect[2] * 255));
svg.svgprint(file_name+'.svg')
def plot_intrinsic(X, w, h):
svg = MYSVGWriter(640, 480, -100, -100, 100, 100)
x_prev, y_prev = None, None
for row in X:
x= 10**1*row[0] #to scale data
y= 10**1*row[1] #to scale data
svg.dot(x ,y,1, 0x008080)
if x_prev and y_prev:
svg.line(x_prev, y_prev, x, y, .05, 0xc0c0c0)
x_prev=x
y_prev=y
svg.svgprint("intrinsic.svg")
if __name__ == '__main__':
matrix, cat_cols, nom_cols = read_data('data/credit-a.arff')
num_folds=3
nfold = NFoldValidation(matrix, cat_cols, nom_cols,num_folds)
print 'total error in linear regression:', nfold.run()
components=5
pca02= PCA02(matrix, cat_cols, nom_cols, components)
pca02.train()
#plot3.svg
svg = MYSVGWriter(500, 500, 0, 0, 100, 100)
start_position=4
for i in range(len( pca02.eigenvalues)):
svg.rect(start_position, 0, 3, 27*pca02.eigenvalues[i] , 0x008080)
start_position+= 4
# print pca02.eigenvalues
svg.svgprint("plot3.svg")
#plot4.svg
svg = MYSVGWriter(500, 500, -100, -100, 110, 110)
start_position=10
for i in range(len( pca02.matrix)):
row = pca02.matrix[i]
reduced= pca02.reduce(row[:len(row)], 2)
x= 10**17*reduced[0]#to scale data
