feature_scoring_flu.py

"""Basic feature scoring example."""

import matplotlib.pyplot as plt
import numpy as np
import seaborn as sns

from ema_workbench import ema_logging, load_results
from ema_workbench.analysis import feature_scoring

ema_logging.log_to_stderr(level=ema_logging.INFO)

# load data
fn = r"./data/1000 flu cases with policies.tar.gz"
x, outcomes = load_results(fn)

# we have timeseries so we need scalars
y = {
    "deceased population": outcomes["deceased_population_region_1"][:, -1],
    "max. infected fraction": np.max(outcomes["infected_fraction_R1"], axis=1),
}

scores = feature_scoring.get_feature_scores_all(x, y)
sns.heatmap(scores, annot=True, cmap="viridis")
plt.show()