How to Create a Bar Chart in Python: matplotlib, seaborn, and plotly

The query "how to create a bar chart in Python" sounds simple, but publication-ready bar charts need more than one line of code. In scientific work, you often need grouped comparisons, uncertainty bars, and clear axis labeling that survives peer review.

1. Basic matplotlib bar chart

Start with plain matplotlib for full control. Use a y-axis starting at zero and annotate bar values for fast reading.

import matplotlib.pyplot as plt

categories = ["Control", "Treatment A", "Treatment B", "Treatment C"]
values = [62.4, 71.2, 79.8, 68.9]

fig, ax = plt.subplots(figsize=(8, 5))
ax.bar(categories, values, color=["#4f46e5", "#0891b2", "#16a34a", "#d97706"])
ax.set_ylabel("Response score")
ax.set_xlabel("Condition")
ax.set_title("Mean response by condition")
ax.set_ylim(0, 90)
ax.grid(axis="y", alpha=0.25)

for i, value in enumerate(values):
    ax.text(i, value + 1.2, f"{value:.1f}", ha="center", fontsize=10)

plt.tight_layout()

2. Grouped bar chart in matplotlib

Grouped bars are the standard for treatment-vs-control across time points or assay conditions. Offset x positions using width/2.

import numpy as np
import matplotlib.pyplot as plt

conditions = ["Week 1", "Week 2", "Week 3", "Week 4"]
control = [55, 57, 60, 62]
treatment = [56, 62, 69, 74]

x = np.arange(len(conditions))
width = 0.38

fig, ax = plt.subplots(figsize=(9, 5))
ax.bar(x - width / 2, control, width, label="Control", color="#64748b")
ax.bar(x + width / 2, treatment, width, label="Treatment", color="#10b981")

ax.set_xticks(x)
ax.set_xticklabels(conditions)
ax.set_ylabel("Cell viability (%)")
ax.set_title("Grouped bar chart across time")
ax.legend(frameon=False)
ax.grid(axis="y", alpha=0.25)

plt.tight_layout()

3. Bar chart with error bars

For scientific plots, pass yerr and capsize. Then state in your caption whether bars represent SD, SEM, or 95% CI.

import numpy as np
import matplotlib.pyplot as plt

groups = ["Control", "Drug A", "Drug B", "Drug C"]
means = np.array([1.00, 1.22, 1.41, 1.18])
sem = np.array([0.06, 0.07, 0.09, 0.05])

fig, ax = plt.subplots(figsize=(8, 5))
ax.bar(groups, means, yerr=sem, capsize=5, color="#0f172a", alpha=0.88)
ax.set_ylabel("Normalized expression")
ax.set_title("Bar chart with SEM error bars")
ax.set_ylim(0, 1.65)
ax.grid(axis="y", alpha=0.25)

for i, mean in enumerate(means):
    ax.text(i, mean + sem[i] + 0.03, f"{mean:.2f}", ha="center", fontsize=10)

plt.tight_layout()

4. Seaborn vs plotly vs matplotlib

Use matplotlib when you need low-level layout control, seaborn when you want fast statistical defaults, and plotly when interactivity matters.

Seaborn quick start

import seaborn as sns
import matplotlib.pyplot as plt

# df columns: condition, value
sns.set_theme(style="whitegrid")
ax = sns.barplot(data=df, x="condition", y="value", errorbar=("ci", 95), palette="crest")
ax.set_title("Seaborn barplot with 95% CI")
ax.set_xlabel("Condition")
ax.set_ylabel("Signal intensity")
plt.tight_layout()

Plotly quick start

import plotly.express as px

# df columns: condition, value, sem
fig = px.bar(
    df,
    x="condition",
    y="value",
    error_y="sem",
    color="condition",
    title="Interactive bar chart with error bars",
)
fig.update_layout(template="plotly_white", showlegend=False)
fig.show()

Common bar chart mistakes

• Non-zero baseline - truncating y-axis can exaggerate small differences.
• Missing uncertainty - no error bars makes interpretation weak for reviewer scrutiny.
• Too many categories - for long labels or 15+ groups, use horizontal bars or a dot plot.
• Rainbow palette overload - use restrained, purposeful colors tied to conditions.