Python Ellipse.set_fill方法代码示例

# 需要导入模块: from matplotlib.patches import Ellipse [as 别名]
# 或者: from matplotlib.patches.Ellipse import set_fill [as 别名]
def covarianceEllipse(mean, covar, scale=1.0):
    eval, evec = np.linalg.eig(covar)
    angles = np.degrees(np.arctan(evec[1,:] / evec[0,:]))
    q1 = np.flatnonzero((angles >= 0) & (angles < 90))[0]
    q2 = 1 - q1
    q1Angle = angles[q1]
    q1Length = np.sqrt(eval[q1])
    q2Length = np.sqrt(eval[q2])
    e = Ellipse(xy=mean, angle=q1Angle,
                width=q1Length*scale, height=q2Length*scale)
    e.set_fill(False)
    e.set_edgecolor("black")
    e.set_alpha(1.0)
    e.set_lw(2)
    e.set_ls("dashed")
    return e

# 需要导入模块: from matplotlib.patches import Ellipse [as 别名]
# 或者: from matplotlib.patches.Ellipse import set_fill [as 别名]
    def ellipses2D(self, colors):
        from matplotlib.patches import Ellipse

        ellipses = []

        for i, ((weight, mean, _), covar) in enumerate(zip(self, self._get_covars())):
            (val, vect) = eig(covar)

            el = Ellipse(mean,
                         3.5 * sqrt(val[0]),
                         3.5 * sqrt(val[1]),
                         180. * arctan2(vect[1, 0], vect[0, 0]) / pi,
                         fill=False,
                         linewidth=2)

            el.set_facecolor(colors[i])
            el.set_fill(True)
            el.set_alpha(0.5)

            ellipses.append(el)

        return ellipses

# 需要导入模块: from matplotlib.patches import Ellipse [as 别名]
# 或者: from matplotlib.patches.Ellipse import set_fill [as 别名]
def plotCovarianceCorner(covariance,labels=None,percentiles=[67,95,99.7],alphas=[1.,0.7,0.5],addToExisitingAxes=None,fillEllipses=True,
						univariateAlpha=1.,ellipseFaceColour='k',ellipseEdgeColour='k',univariateColour='k',maxStds=None,
						means=None,labelSize=20,tickSize=20):

	nVariate = len(covariance)

	if labels == None:
		labels = ['' for _ in range(nVariate)]

	## work out the critical values according to the provided percentiles
	criticals = []
	for p in percentiles:
		criticals.append(spst.chi2.ppf(q=p/100.,df=2))

	#plot in descending percentile (smaller ellipses on top)
	criticals = np.sort(criticals)
	alphas = np.sort(alphas)[::-1]

	stds = np.sqrt(np.diag(covariance))

	#sensible widths for plots
	if np.any(maxStds == None):
		widths = stds * np.max(np.sqrt(criticals)) * 1.2
	else:
		widths = maxStds * np.max(np.sqrt(criticals)) * 1.2

	widths = np.array(widths)

	if means is None:
		means = np.zeros_like(np.diag(covariance))

	if np.any(addToExisitingAxes == None):
		fig,axs = plot.subplots(nVariate,nVariate)
	else:
		fig,axs = addToExisitingAxes

	# for i in range(nVariate):
	# 	for j in range(nVariate):
	for i,j in np.ndindex(nVariate,nVariate):

			if j>i:
				#hide upper right triangle
				axs[i,j].axis('off')

			elif i == j:

				#univariate plots
				var = covariance[i,j]

				xs = np.linspace(-widths[i], widths[i],1000) + means[i]
				# ys = np.exp(-0.5*((xs-means[i])**2)/var)
				ys = spst.norm.pdf(xs,loc=means[i],scale=np.sqrt(var))
				# ys = ys/np.max(ys)

				axs[i,j].plot(xs,ys,color=univariateColour,alpha=univariateAlpha)

				axs[i,j].set_xlim(-widths[i]+means[i], widths[i]+means[i])

				# axs[i,j].set_ylim(0,1.1)

				axs[i,j].set_yticks([])

			else:
				#bivariate plots
				bivCov = np.array([[covariance[i,i],covariance[i,j]],
								   [covariance[j,i],covariance[j,j]]])

				eigVals, eigVecs = np.linalg.eig(bivCov)

				princEigVec = eigVecs[:,0]

				angle = np.arctan(princEigVec[0]/princEigVec[1])*(180./np.pi)
				#np.arctan2()
				#np.rad2deg()

				for en,c in enumerate(criticals):

					e = Ellipse((means[j],means[i]), 2.*np.sqrt(c*eigVals[0]), 2.*np.sqrt(c*eigVals[1]),angle=angle)
					e.set_alpha(alphas[en])
					e.set_fill(fillEllipses)
					e.set_edgecolor(ellipseEdgeColour)
					e.set_facecolor(ellipseFaceColour)

					axs[i,j].add_artist(e)
					axs[i,j].set_xlim(-widths[j]+means[j],widths[j]+means[j])
					axs[i,j].set_ylim(-widths[i]+means[i],widths[i]+means[i])

					#don't overcrowd the axes
					axs[i,j].set_xticks(np.linspace(-misc.roundToOneSigFig(widths[j]*0.5),misc.roundToOneSigFig(widths[j]*0.5),3) + means[j])
					axs[i,j].set_yticks(np.linspace(-misc.roundToOneSigFig(widths[i]*0.5),misc.roundToOneSigFig(widths[i]*0.5),3) + means[i])


			#only label the left most and bottom most plots
			if i>0 and j == 0:
				axs[i,j].set_ylabel(labels[i],size=labelSize)
				for tick in axs[i,j].yaxis.get_major_ticks():
					tick.label1.set_fontsize(tickSize)

			if i == nVariate - 1:
				axs[i,j].set_xlabel(labels[j],size=labelSize)
#.........这里部分代码省略.........

# 需要导入模块: from matplotlib.patches import Ellipse [as 别名]
# 或者: from matplotlib.patches.Ellipse import set_fill [as 别名]
    # for GMM
    # covar = clf.covars_[i][0:2, 0:2]

    # for all modes of GMM
    covar = clf._get_covars()[i][0:2, 0:2]

    v, w, = linalg.eigh(covar)
    angle = np.arctan2(w[0][1], w[0][0]) * 180.0 / np.pi
    angle_out[i] = angle

    ell = Ellipse(means, v[0], v[1], angle)
    ax.add_artist(ell)
    ell.set_clip_box(ax.bbox)
    ell.set_facecolor('None')
    ell.set_fill(False)
    ell.set_color(color_iter.next())
    # ell.set_alpha(.25)

plt.xlim((np.min(tmid), np.max(tmid)))
plt.ylim((0,360))
plt.scatter(xo, yo, c='k', s=(zo / np.nanmax(r1)*20.)**2., alpha=0.6)
plt.show()

plt.figure()
h = plt.hist(angle_out)
plt.show()