Introduction to Distributions¶

When you have a few chains and want to contrast them all with each other, you probably want a summary plot.

To show you how they work, let's make some sample data that all has the same average.

from chainconsumer import Chain, ChainConsumer, Truth, make_sample

# Here's what you might start with
df_1 = make_sample(num_dimensions=4, seed=1, randomise_mean=True)
df_2 = make_sample(num_dimensions=5, seed=2, randomise_mean=True)
print(df_1.head())

Out:

          A         B          C          D  log_posterior
0 -0.668065  4.875194   8.637756  15.725842      -3.216375
1 -2.484174  6.123240  10.946138  13.550658      -4.115756
2 -1.300656  5.926036  10.044298  14.147247      -2.621589
3 -0.102935  3.067071   8.375315  14.681300      -5.474791
4 -1.949402  5.115333  11.353940  12.323536      -4.419287

Using distributions¶

# And now we give this to chainconsumer
c = ChainConsumer()
c.add_chain(Chain(samples=df_1, name="An Example Contour"))
fig = c.plotter.plot_distributions()

$A = -0.71^{+0.99}_{-1.02}$, $B = 4.81^{+1.00}_{-1.01}$, $C = 9.43^{+0.98}_{-1.03}$, $D = 14.55^{+0.99}_{-1.03}$

If you want the summary stats you'll need to keep it just one chain. And if you don't want them, you can pass summarise=False to the PlotConfig.

When you add a second chain, you'll see the summaries disappear.

c.add_chain(Chain(samples=df_2, name="Another contour!"))
c.add_truth(Truth(location={"A": 0, "B": 0}))
fig = c.plotter.plot_distributions(col_wrap=3, columns=["A", "B"])