What does your model believe?

Drag your prior distributions. Simulate their consequences. Keep the code.

Your model

Python · PyMC

Paste a self-contained model named model. Include your inputs; leave out fitting calls.

import numpy as np
import pymc as pm

x = np.linspace(-1, 1, 60)

with pm.Model() as model:
    intercept = pm.Normal("intercept", mu=0, sigma=1)
    slope = pm.Normal("slope", mu=0, sigma=2)
    noise = pm.Gamma("noise", alpha=2, beta=2)
    mu = pm.Deterministic("mu", intercept + slope * x)
    y = pm.Normal("y", mu=mu, sigma=noise,
                  observed=np.zeros(len(x)))

How code and controls stay in sync

Modist curves and numeric controls edit numeric literals in your original Python. Expressions and dependent parameters stay in code. Read priors inspects syntax without executing your model. Simulate runs the full code locally; only run code you trust, which can access the network. The first run loads about 120 MB of Python assets.

No data is fitted. Observed values define the prediction shape; their values do not condition the prior. Changes last for this page session. Export to keep your work.

Prior assumptions

↔ code

interceptNormal

sigma

slopeNormal

sigma

noiseGamma

alpha

beta

DrawsSeed

Ready to explore

Prior predictive

Not yet sampled

Meet the data your priors imply.

Start with a model on the left. Simulate to see its possible outcomes before seeing any evidence.

01 Inspect the assumptions02 Draw possible observations03 Check what feels plausible