Changing a `suggest_uniform` call trigger an error in plot_param_importances when `params` argument given.

If the database (I use the potgresql backend for storing trials) contains entries:

598|61|x|51.0|{"name": "IntUniformDistribution", "attributes": {"low": 20, "high": 100, "step": 1}}
608|62|x|63.0|{"name": "IntUniformDistribution", "attributes": {"low": 30, "high": 120, "step": 1}}

then using suggest_uniform(study, params=['x']) raise a ValueError: Parameters importances cannot be assessed with dynamic search spaces if parameters are specified. Specified parameters: ['x']..

I have to manually edit the entries in the database to low:20, high:120 to fix the issue. 😦

It would be nice if the plotting function could do this kumfu by itself 😃

[hvy]

I’m wondering if this previous discussion doesn’t address the topic here https://github.com/optuna/optuna/issues/1856#issuecomment-703935388. A PR was worked on but unfortunately got stale.

[nzw0301]

Thank you for creating the issue! According to the error message, I suppose that the behaviour is intended, so let me change the label to feature from bug.

At least, it would be great to give a note to explain the dynamical range is not supported by plot_param_importances.

---

Let me put a minimal example to see the error message.

import optuna


def objective(trial):
    x = trial.suggest_int("x", 0, trial.number)
    y = trial.suggest_float("y", -1.0, 1.0)
    z = trial.suggest_float("z", 0.0, 1.5)
    return x ** 2 + y ** 3 - z ** 4


sampler = optuna.samplers.RandomSampler(seed=10)
study = optuna.create_study(sampler=sampler)
study.optimize(objective, n_trials=10)

optuna.visualization.plot_param_importances(study, params=["x"])