BUG: Waterfall feature names IndexError

[LukeHankey]

Issue Description

When a shapley explainer is given an nxn feature matrix, the waterfall plot fails to be created because the feature names are not a list of the n feature names, but only the single feature at the sample position given.

The Explanation object has this code in __init__:

        if len(_compute_shape(feature_names)) == 1: # TODOsomeday: should always be an alias once slicer supports per-row aliases
            if len(values_shape) >= 1 and len(feature_names) == values_shape[0]:
                feature_names = Alias(list(feature_names), 0)
            elif len(values_shape) >= 2 and len(feature_names) == values_shape[1]:
                feature_names = Alias(list(feature_names), 1)

The Alias dim of the feature_names is set to 0 when the features and samples match which results in just the single feature "Feature_1" instead of the list of features "['Feature_1', 'Feature_2', 'Feature_x']". The Explanation object then saves the Slicer object in self._s which when the waterfall plot is created with just a single sample shap_explainer[0], __get_item__ is called and then the slice uses the dim from above to either select the index given (in this case 0) when dim=0, or slice(None, None, None) for dim=1.

Minimal Reproducible Example

import pandas as pd
import numpy as np
import shap
import xgboost

model = xgboost.XGBClassifier()
n_samples, n_features = 400, 30
df = pd.DataFrame(np.random.random((n_samples, n_features)), columns=[f"Feature_{i}" for i in range(n_features)])

model.fit(df, np.random.choice([0, 1], n_samples))

explainer = shap.TreeExplainer(model)
shap_explainer = explainer(df.sample(n_features))

shap.plots.waterfall(shap_explainer[0])

Traceback

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/luke/oxcan-fs/venv/lib/python3.10/site-packages/shap/plots/_waterfall.py", line 139, in waterfall
    yticklabels[rng[i]] = format_value(float(features[order[i]]), "%0.03f") + " = " + feature_names[order[i]]
IndexError: string index out of range

Expected Behavior

The slice should always get the full set of features when creating a waterfall plot.

Bug report checklist

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest release of shap.
• I have confirmed this bug exists on the master branch of shap.
• I'd be interested in making a PR to fix this bug

Installed Versions

0.42.1

[CloseChoice]

I cannot reproduce this. This works for me on the latest master.