
数据:
脚本:
from pymc3 import *from numpy.random import uniform,normalbp_u = 30 #switch pointc_u = [1,-1] #intercepts before and after switch pointbeta_u = [0,-0.02] #slopes before & after switch pointx = uniform(0,90,200)y = (x < bp_u)*(c_u[0]+beta_u[0]*x) + (x >= bp_u)*(c_u[1]+beta_u[1]*x) + normal(0,0.1,200)with Model() as sw_model: sigma = HalfCauchy('sigma',beta=10,testval=1.) switchpoint = Uniform('switchpoint',lower=x.min(),upper=x.max(),testval=45) # Priors for pre- and post-switch intercepts and slopes intercept_u1 = Uniform('Intercept_u1',lower=-10,upper=10) intercept_u2 = Uniform('Intercept_u2',upper=10) x_coeff_u1 = normal('x_u1',sd=20) x_coeff_u2 = normal('x_u2',sd=20) intercept = switch(switchpoint < x,intercept_u1,intercept_u2) x_coeff = switch(switchpoint < x,x_coeff_u1,x_coeff_u2) likelihood = normal('y',mu=intercept + x_coeff * x,sd=sigma,observed=y) start = find_MAP() with sw_model: step1 = NUTS([intercept_u1,intercept_u2,x_coeff_u2]) step2 = NUTS([switchpoint]) trace = sample(2000,step=[step1,step2],start=start,progressbar=True) 以下是结果:
如您所见,它们与初始值完全不同.我做错了什么?
解决方法 最后,似乎用Metropolis采样切换到离散断点解决了这个问题.这是最终的模型:with Model() as sw_model: sigma = HalfCauchy('sigma',testval=1.) switchpoint = discreteUniform('switchpoint',lower=0,upper=90,upper=10,testval = 0) intercept_u2 = Uniform('Intercept_u2',testval = 0) x_coeff_u1 = normal('x_u1',observed=y) start = find_MAP() step1 = NUTS([intercept_u1,x_coeff_u2]) step2 = Metropolis([switchpoint]) trace = sample(20000,njobs=4,progressbar=True) 总结 以上是内存溢出为你收集整理的python – 具有变化点的PyMC3回归全部内容,希望文章能够帮你解决python – 具有变化点的PyMC3回归所遇到的程序开发问题。
如果觉得内存溢出网站内容还不错,欢迎将内存溢出网站推荐给程序员好友。
欢迎分享,转载请注明来源:内存溢出
微信扫一扫
支付宝扫一扫
评论列表(0条)