Simulating choices¶

Often we will want to create simulated data containing choices made based on the values learned by a particular model.

In [1]:

from behavioural_modelling.utils import choice_from_action_p
import jax
import numpy as np
import jax.numpy as jnp

from behavioural_modelling.utils import choice_from_action_p import jax import numpy as np import jax.numpy as jnp

Example of generating choices based on choice probabilities¶

We can use the choice_from_action_p function to generate choices based on the action probabilities output by a model.

As this is built using JAX, it requires a random key to be passed in. This can be generated using the jax.random.PRNGKey function.

In [2]:

key = jax.random.PRNGKey(0)

key = jax.random.PRNGKey(0)

WARNING:jax._src.lib.xla_bridge:No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)

We can then generate some action probabilities.

In [3]:

rng = np.random.default_rng(12345)

# Generate random probabilities
probs = rng.uniform(size=(40, 100, 3))

rng = np.random.default_rng(12345) # Generate random probabilities probs = rng.uniform(size=(40, 100, 3))

And generate choices.

In [8]:

choices = choice_from_action_p(key, probs)
print(choices)

choices = choice_from_action_p(key, probs) print(choices)

[[2 1 2 ... 2 1 2]
 [1 0 1 ... 1 1 1]
 [1 2 1 ... 1 1 2]
 ...
 [2 1 1 ... 1 1 1]
 [0 1 1 ... 1 2 1]
 [1 2 1 ... 1 0 2]]