Alternating Ribbon Fill with Disjoint Groups

This notebook demonstrates a subtle issue that arises when using geom_ribbon() to visualize piecewise geometry with repeated fill categories.

from math import sin

import pandas as pd

from lets_plot import *

LetsPlot.setup_html()

The setup involves two sine waves with a phase shift, producing alternating intersection regions. We want to highlight the area between the curves using alternating colors (e.g., red, blue, red, blue…).

However, if only two fill categories are used (e.g., "a" and "b"), Lets-Plot will treat each category as a single continuous area. As a result, visually disconnected regions with the same category will be merged — which creates unwanted bridging between non-adjacent segments.

def get_data():
    x = list(range(60))
    a = [sin(v / 10.0) * 20.0 for v in x]
    b = [sin(v / 6.0 + 1.0) * 20.0 - 2.0 for v in x]
    return pd.DataFrame({
        "x": x,
        "min": [min(first, second) for (first, second) in zip(a, b)],
        "max": [max(first, second) for (first, second) in zip(a, b)],
        "label": [("a" if first > second else "b") for (first, second) in zip(a, b)],
    })

df = get_data()
df.head(10)

	x	min	max	label
0	0	0.000000	14.829420	b
1	1	1.996668	16.388900	b
2	2	3.973387	17.438758	b
3	3	5.910404	17.949900	b
4	4	7.788367	17.908159	b
5	5	9.588511	17.314693	b
6	6	11.292849	16.185949	b
7	7	12.884354	14.553207	b
8	8	12.461718	14.347122	a
9	9	9.969443	15.666538	a

ggplot(df) + \
    geom_ribbon(aes(x="x", ymin="min", ymax="max", fill="label", color="label"), alpha=.2)

To fix this, we generate unique group labels for each individual segment (e.g., "a1", "a2", "a3", "a4", …). During plotting, we assign the same color to all even segments and another to odd segments using scale_manual(). This ensures proper coloring while keeping the regions visually separated.

def update_data(df):
    return df.assign(
        label="a" + df["label"].ne(df["label"].shift()).cumsum().astype(str)
    )

corrected_df = update_data(df)
corrected_df.head(10)

	x	min	max	label
0	0	0.000000	14.829420	a1
1	1	1.996668	16.388900	a1
2	2	3.973387	17.438758	a1
3	3	5.910404	17.949900	a1
4	4	7.788367	17.908159	a1
5	5	9.588511	17.314693	a1
6	6	11.292849	16.185949	a1
7	7	12.884354	14.553207	a1
8	8	12.461718	14.347122	a2
9	9	9.969443	15.666538	a2

ggplot(corrected_df) + \
    geom_ribbon(aes(x="x", ymin="min", ymax="max", fill="label", color="label"), alpha=.2) + \
    scale_manual(["color", "fill"], values=["#e41a1c", "#377eb8"], breaks=[1, 2], labels=["b", "a"])

Additionally, we modify the dataset to eliminate visible gaps between categories by inserting extra points at the segment boundaries.

def get_continuous_df(df, target_col):
    cols = df.columns
    out_rows = []
    for i in range(len(df)):
        if i > 0 and df[target_col].iloc[i] != df[target_col].iloc[i-1]:
            boundary = df.iloc[i].copy()
            boundary[target_col] = df[target_col].iloc[i-1]
            out_rows.append(boundary)
        out_rows.append(df.iloc[i])
    return pd.DataFrame(out_rows, columns=cols).reset_index(drop=True)

continuous_df = get_continuous_df(corrected_df, "label")
continuous_df.head(10)

	x	min	max	label
0	0	0.000000	14.829420	a1
1	1	1.996668	16.388900	a1
2	2	3.973387	17.438758	a1
3	3	5.910404	17.949900	a1
4	4	7.788367	17.908159	a1
5	5	9.588511	17.314693	a1
6	6	11.292849	16.185949	a1
7	7	12.884354	14.553207	a1
8	8	12.461718	14.347122	a1
9	8	12.461718	14.347122	a2

ggplot(continuous_df) + \
    geom_ribbon(aes(x="x", ymin="min", ymax="max", fill="label", color="label"), alpha=.2) + \
    scale_manual(["color", "fill"], values=["#e41a1c", "#377eb8"], breaks=[1, 2], labels=["b", "a"])