我们只使用SailsJS作为JSON REST服务器，没有html前端。但是我们必须定义要输出的JSON格式。
正如我目前所见，有两种方法可以指定json格式：
import pymc3 as pm
import numpy as np
import matplotlib.pyplot as plt
import multiprocessing
import seaborn as sns
import pandas as pd
import theano.tensor as tt
%matplotlib inline
# Clip at 15 components
K = 15
# Create mixture population
centroids = [0, 10, 50]
weights = [(2/5),(2/5),(1/5)]
mix_3 = np.concatenate([np.random.normal(loc=centroids[0], size=int(150*weights[0])), # 60 samples
np.random.normal(loc=centroids[1], size=int(150*weights[1])), # 60 samples
np.random.normal(loc=centroids[2], size=int(150*weights[2]))])# 30 samples
n = mix_3.size
# Create and fit model
with pm.Model() as Mod_dir:
alpha = pm.Gamma('alpha', 1., 1.)
beta = pm.Beta('beta', 1., alpha, shape=K)
w = pm.Deterministic('w', beta * tt.concatenate([[1], tt.extra_ops.cumprod(1 - beta)[:-1]]))
component = pm.Categorical('component', w, shape=n)
tau = pm.Gamma("tau", 1.0, 1.0, shape=K)
mu = pm.Normal('mu', 0, tau=tau, shape=K)
obs = pm.Normal('obs',
mu[component],
tau=tau[component],
observed=mix_3)
step1 = pm.Metropolis(vars=[alpha, beta, w, tau, mu, obs])
# step2 = pm.CategoricalGibbsMetropolis(vars=[component])
step2 = pm.ElemwiseCategorical([component], np.arange(K)) # Much, much faster than the above
tr = pm.sample(1e4, [step1, step2], njobs=multiprocessing.cpu_count())
#burn-in = 1000, thin by grabbing every 5th idx
pm.traceplot(tr[1e3::5])
和module.exports.attributes.toJSON()定义不同的格式。我相信，正确的方法是使用像HTML前端的视图，但是使用JSON。有没有现成的解决方案？