e8b2e17f4eec658f6a6d53486dbbe3eac48bb0fa,pyinterpret/core/global_interpretation/partial_dependence.py,PartialDependence,_plot_2d_2_binary_feature,#PartialDependence#Any#Any#Any#Any#Any#Any#,619

Before Change

            for val in np.unique(pdp[feature2]):
                color = next(colors)
                filter_idx = pdp[feature2] == val
                pdp_vals = pdp[filter_idx][mean_col].values
                x1 = pdp[filter_idx][feature1].values
                ax.bar(x1, pdp_vals, color=color,
                       label="{} = {}".format(*[feature2, val], align="center")

After Change

                                  class_col_pairs, with_variance=False):
        figure_list, axis_list = [], []
        sd_col = pdp_metadata["sd_col"]
        std_error = pdp.set_index([feature1, feature2])[sd_col].unstack()
        for class_name, mean_col in class_col_pairs:
            f = plt.figure()
            ax = f.add_subplot(111)
            //feature2 is columns
            //feature1 is index
            plot_data = pdp.set_index([feature1, feature2])[mean_col].unstack()
            plot_data.plot(ax=ax, color=COLORS)

            if with_variance:
                colors = cycle(COLORS)
                binary1_values = plot_data.index.values
                binary2_values = plot_data.columns.values
                for binary2_value in binary2_values:
                    color = next(colors)
                    yerr = std_error[binary2_value].values
                    upper_plane = yerr + plot_data[binary2_value].values
                    lower_plane = plot_data[binary2_value].values - yerr
                    ax.fill_between(binary1_values, lower_plane, upper_plane,
                                    color=color,alpha=.2)
            figure_list.append(f)
            axis_list.append(ax)
            ax.set_xlabel(feature1)
            ax.set_ylabel("Predicted {}".format(class_name))

Instances