mlcolvar.explain.sensitivity.plot_sensitivity¶

class mlcolvar.explain.sensitivity.plot_sensitivity(results, mode='violin', per_class=None, max_features=100, ax=None)[source]¶

Bases:

Plot results of the sensitivity analysis. They can be plotted in three modes: * Violin plot (‘violin’), showing the density of per-sample sensitivities besides the mean value * Scatter (‘scatter’), plotting the mean and standard deviation of the gradients * Horizontal bar plot (‘barh’) only displaying the mean of the gradients

Parameters:

results (dictionary) – sensitity results calculated by sensitivity_analysis
mode (string, optional) – (‘violin’,’barh’,’scatter’), by default ‘violin’
per_class (bool, optional) – plot per-class statistics if available, by default plot them if available
max_features (int, optional) – plot at most max_features, by default 100
ax (matplotlib axis, optional) – ax where to plot the results, by default it will be initialized