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: matplotlib 3.10.9 | Python 3.13.13
Quality: 90/100 | Created: 2026-05-07
"""
import os
import matplotlib.patches as mpatches
import matplotlib.pyplot as plt
import numpy as np
# Theme tokens
THEME = os.getenv("ANYPLOT_THEME", "light")
PAGE_BG = "#FAF8F1" if THEME == "light" else "#1A1A17"
ELEVATED_BG = "#FFFDF6" if THEME == "light" else "#242420"
INK = "#1A1A17" if THEME == "light" else "#F0EFE8"
INK_SOFT = "#4A4A44" if THEME == "light" else "#B8B7B0"
INK_MUTED = "#6B6A63" if THEME == "light" else "#A8A79F"
# Semantic direction colors — imprint diverging anchors
COLOR_POS = "#AE3030" # red: positive SHAP (pushes prediction up)
COLOR_NEG = "#4467A3" # blue: negative SHAP (pushes prediction down)
# Data — credit loan approval model, individual applicant explanation
# Features ordered by absolute SHAP magnitude, largest first (top of chart)
features_desc = [
"Monthly Income",
"Debt-to-Income Ratio",
"Credit Score",
"Employment Duration",
"Prior Defaults",
"Loan Amount",
"Age",
"Open Credit Lines",
"Savings Balance",
"Recent Inquiries",
]
shap_desc = [0.24, -0.18, 0.15, 0.12, -0.10, -0.08, 0.06, 0.04, 0.03, -0.02]
base_value = 0.42 # E[f(x)]: mean approval probability across training set
final_value = base_value + sum(shap_desc) # f(x) = 0.68
# Reverse for matplotlib bottom-to-top axis (largest magnitude at top)
features = features_desc[::-1]
shap_values = shap_desc[::-1]
n = len(features)
y_pos = np.arange(n)
# Compute bar left edges and cumulative running totals (bottom-to-top)
running = base_value
bar_lefts = []
cum_after = []
for sv in shap_values:
bar_lefts.append(running + sv if sv < 0 else running)
running += sv
cum_after.append(running)
bar_widths = [abs(sv) for sv in shap_values]
bar_colors = [COLOR_POS if sv >= 0 else COLOR_NEG for sv in shap_values]
# Plot
fig, ax = plt.subplots(figsize=(16, 9), facecolor=PAGE_BG)
ax.set_facecolor(PAGE_BG)
ax.set_axisbelow(True)
ax.barh(y_pos, bar_widths, left=bar_lefts, color=bar_colors, height=0.55, edgecolor=PAGE_BG, linewidth=1.5, zorder=3)
# Vertical connector lines in the gaps between bars at cumulative position
for i in range(n - 1):
ax.plot(
[cum_after[i], cum_after[i]],
[y_pos[i] + 0.32, y_pos[i + 1] - 0.32],
color=INK_MUTED,
linewidth=1.2,
linestyle="dotted",
alpha=0.7,
zorder=2,
)
# Reference lines for base value and final prediction
ax.axvline(base_value, color=INK_MUTED, linewidth=1.5, linestyle="--", alpha=0.6)
ax.axvline(final_value, color=INK_SOFT, linewidth=2.0, linestyle="--", alpha=0.7)
# SHAP value text labels on (wide bars) or beside (narrow bars) each segment
LABEL_THRESH = 0.07
for i, (sv, left, width) in enumerate(zip(shap_values, bar_lefts, bar_widths, strict=False)):
txt = f"+{sv:.2f}" if sv > 0 else f"{sv:.2f}"
if width >= LABEL_THRESH:
ax.text(
left + width / 2,
y_pos[i],
txt,
ha="center",
va="center",
fontsize=14,
fontweight="bold",
color="white",
zorder=5,
)
else:
gap = 0.007
if sv > 0:
ax.text(
left + width + gap,
y_pos[i],
txt,
ha="left",
va="center",
fontsize=14,
fontweight="bold",
color=COLOR_POS,
zorder=5,
)
else:
ax.text(
left - gap,
y_pos[i],
txt,
ha="right",
va="center",
fontsize=14,
fontweight="bold",
color=COLOR_NEG,
zorder=5,
)
# Annotations for base value and final prediction above the top bar
top_y = n + 0.08
ax.text(
base_value,
top_y,
f"E[f(x)] = {base_value:.2f}",
ha="center",
va="bottom",
fontsize=13,
color=INK_MUTED,
bbox={"facecolor": ELEVATED_BG, "edgecolor": INK_MUTED, "boxstyle": "round,pad=0.35", "alpha": 0.9},
)
ax.text(
final_value,
top_y,
f"f(x) = {final_value:.2f}",
ha="center",
va="bottom",
fontsize=13,
fontweight="bold",
color=INK,
bbox={"facecolor": ELEVATED_BG, "edgecolor": INK_SOFT, "boxstyle": "round,pad=0.35", "alpha": 0.9},
)
# Style
ax.set_yticks(y_pos)
ax.set_yticklabels(features, fontsize=16)
ax.set_xlabel("Loan Approval Probability", fontsize=20, color=INK)
ax.set_title("Credit Approval · shap-waterfall · matplotlib · anyplot.ai", fontsize=24, fontweight="medium", color=INK)
all_x = bar_lefts + [b + w for b, w in zip(bar_lefts, bar_widths, strict=False)] + [base_value, final_value]
ax.set_xlim(min(all_x) - 0.06, max(all_x) + 0.09)
ax.set_ylim(-0.6, n + 0.9)
ax.tick_params(axis="x", labelsize=16, colors=INK_SOFT)
ax.tick_params(axis="y", labelsize=16, colors=INK_SOFT, length=0)
ax.spines["top"].set_visible(False)
ax.spines["right"].set_visible(False)
ax.spines["left"].set_visible(False)
ax.spines["bottom"].set_color(INK_SOFT)
ax.xaxis.grid(True, alpha=0.10, linewidth=0.8, color=INK)
# Legend
pos_patch = mpatches.Patch(color=COLOR_POS, label="Positive contribution")
neg_patch = mpatches.Patch(color=COLOR_NEG, label="Negative contribution")
leg = ax.legend(handles=[pos_patch, neg_patch], fontsize=16, loc="lower right", frameon=True)
leg.get_frame().set_facecolor(ELEVATED_BG)
leg.get_frame().set_edgecolor(INK_SOFT)
plt.setp(leg.get_texts(), color=INK_SOFT)
plt.tight_layout()
plt.savefig(f"plot-{THEME}.png", dpi=300, bbox_inches="tight", facecolor=PAGE_BG)
Part of SHAP Waterfall Plot for Feature Attribution on anyplot.ai.