Trying to fit a poisson process



In [1]:

    
%matplotlib inline
from science import *
from scipy.stats import poisson



In [2]:

    
x=np.random.poisson(lam=10,size=1000)



In [3]:

    
def lnprior(mu):
    if 0<=mu<=100:
        return 0.0
    return -np.inf



In [4]:

    
def lnlike(data,mu):
    return log(mu)*sum(data)-mu*len(data)



In [5]:

    
model=MCMCModel2(x,lnprior,lnlike,
                mu=Uniform(0,100))



In [6]:

    
model.run_mcmc(500)
model.plot_chains()









    



Sampling Prior...
Done.
0.13 s
Running MCMC...
Done.
0.59 s






    





<matplotlib.figure.Figure at 0x10b487a50>



In [7]:

    
model.plot_distributions()

What if my data were 2 or 3 data points?



In [21]:

    
x=np.random.poisson(lam=10,size=1)
print x



In [22]:

    
model=MCMCModel2(x,lnprior,lnlike,
                mu=Uniform(0,100))



In [23]:

    
model.run_mcmc(500)
model.plot_chains()









    



Sampling Prior...
Done.
0.14 s
Running MCMC...
Done.
0.58 s






    





<matplotlib.figure.Figure at 0x10c65da50>



In [24]:

    
model.plot_distributions()



In [ ]: