A waterfall-style chart showing how each feature contributes to pushing a model prediction from a base value (expected model output) to the final predicted value. Horizontal bar segments extend right for positive SHAP values and left for negative SHAP values, stacking cumulatively so the viewer can trace the path from baseline to prediction. This is a core ML explainability visualization for explaining individual predictions, complementing the SHAP summary plot which shows feature effects across many samples.
""" anyplot.ai
shap-waterfall: SHAP Waterfall Plot for Feature Attribution
Library: plotly 6.7.0 | Python 3.13.13
Quality: 90/100 | Created: 2026-05-07
"""
import os
import numpy as np
import plotly.graph_objects as go
# Theme tokens
THEME = os.getenv("ANYPLOT_THEME", "light")
PAGE_BG = "#FAF8F1" if THEME == "light" else "#1A1A17"
INK = "#1A1A17" if THEME == "light" else "#F0EFE8"
INK_SOFT = "#4A4A44" if THEME == "light" else "#B8B7B0"
INK_MUTED = "#6B6A63" if THEME == "light" else "#A8A79F"
GRID = "rgba(26,26,23,0.10)" if THEME == "light" else "rgba(240,239,232,0.10)"
# SHAP semantic colors (Okabe-Ito positions used for semantic encoding)
POS_COLOR = "#AE3030" # imprint red — positive SHAP (pushes prediction up)
NEG_COLOR = "#4467A3" # blue — negative SHAP (pushes prediction down)
REF_COLOR = "#009E73" # brand green — reference bars (base & final)
# Data — credit risk model SHAP waterfall for a single high-risk loan applicant
base_value = 0.35 # E[f(x)]: average predicted default rate across training data
final_value = 0.72 # f(x): model's default probability prediction for this applicant
features = [
"Debt-to-Income Ratio",
"Missed Payments (12 mo)",
"Credit Utilization",
"Credit Score",
"Annual Income",
"Employment Stability",
"Credit History Length",
"Account Diversity",
"Property Ownership",
]
shap_values = [0.18, 0.15, 0.12, -0.09, 0.08, -0.06, 0.06, -0.04, -0.03]
# Sort ascending by |SHAP| so the largest contributor appears nearest the top
# (Plotly horizontal waterfall: first y item = bottom, last y item = top)
order = np.argsort(np.abs(shap_values))
sorted_features = [features[i] for i in order]
sorted_shap = [shap_values[i] for i in order]
# Waterfall layers: base (absolute) → features (relative, small→large) → prediction (total)
y_labels = ["E[f(x)] = 0.35"] + sorted_features + ["f(x) = 0.72"]
x_vals = [base_value] + sorted_shap + [0]
measures = ["absolute"] + ["relative"] * len(sorted_features) + ["total"]
text_vals = [f"{base_value:.2f}"] + [f"{v:+.3f}" for v in sorted_shap] + [f"{final_value:.2f}"]
# Plot
fig = go.Figure(
go.Waterfall(
orientation="h",
measure=measures,
y=y_labels,
x=x_vals,
text=text_vals,
textposition="outside",
textfont=dict(size=17, color=INK),
increasing=dict(marker=dict(color=POS_COLOR, line=dict(color=POS_COLOR, width=0))),
decreasing=dict(marker=dict(color=NEG_COLOR, line=dict(color=NEG_COLOR, width=0))),
totals=dict(marker=dict(color=REF_COLOR, line=dict(color=REF_COLOR, width=0))),
connector=dict(line=dict(color=INK_MUTED, width=1.5, dash="dot")),
cliponaxis=False,
)
)
# Layout
fig.update_layout(
title=dict(
text="Credit Default Risk · shap-waterfall · plotly · anyplot.ai",
font=dict(size=26, color=INK),
x=0.5,
xanchor="center",
),
paper_bgcolor=PAGE_BG,
plot_bgcolor=PAGE_BG,
font=dict(color=INK),
xaxis=dict(
title=dict(text="SHAP Value (impact on predicted default probability)", font=dict(size=22, color=INK)),
tickfont=dict(size=18, color=INK_SOFT),
gridcolor=GRID,
linecolor=INK_SOFT,
zerolinecolor=INK_SOFT,
zerolinewidth=2,
range=[-0.20, 0.95],
),
yaxis=dict(tickfont=dict(size=18, color=INK_SOFT), linecolor=INK_SOFT, showgrid=False),
showlegend=False,
margin=dict(l=240, r=160, t=80, b=80),
)
# Save
fig.write_image(f"plot-{THEME}.png", width=1600, height=900, scale=3)
fig.write_html(f"plot-{THEME}.html", include_plotlyjs="cdn")
Part of SHAP Waterfall Plot for Feature Attribution on anyplot.ai.