Choosing the Right Graph: How to Visualize Your Data in Statistics

In data science and machine learning, visualizing your data is one of the first — and most important — steps. This post shows how to choose the right graph (bar chart, pie chart, histogram) based on the type of data you have.

In this post, we’ll explore:

• How to choose a graph based on your data type
• The difference between bar charts, pie charts, and histograms
• The different shapes of histograms (and what they tell us)

---

📚 This post is part of the "Intro to Statistics" series

🔙 Previously: Descriptive vs Inferential Statistics

🔜 Next: Frequency Tables with Python

---

📋 1. Graphs for Categorical Data (Nominal & Ordinal)

Categorical data includes labels, names, or categories.

There are two types:

• Nominal: No order (e.g., eye color, favorite food)
• Ordinal: Ordered categories (e.g., rating from poor to excellent)

🔹 Best graphs for categorical data:

Bar Chart

• Each category is a separate bar
• Bar height = frequency
• Bars are separated (not touching)

Python Example: Bar Chart

import matplotlib.pyplot as plt

labels = ['A', 'B', 'C', 'D', 'E']
sizes = [10, 15, 7, 12, 9]
plt.pie(sizes, labels=labels, autopct='%1.1f%%', startangle=90)
plt.title('Pie Chart Example')
plt.show()

---

Pie Chart

• Shows parts of a whole
• Best when you want to show percentages or proportions

Python Example: Pie Chart

import matplotlib.pyplot as plt

labels = ['A', 'B', 'C', 'D', 'E']
sizes = [10, 15, 7]
plt.pie(sizes, labels=labels, autopct='%1.1f%%', startangle=90)
plt.title('Pie Chart Example')
plt.show()

💡 Tip: Bar charts are usually easier to read than pie charts — especially with many categories.

---

📊 2. Quantitative Data: Interval and Ratio

Quantitative data is numerical, like height, age, or test scores.

This includes:

• Interval: No true zero (e.g., temperature in °C)
• Ratio: Has a true zero (e.g., weight, age)

🔸 Best graph for quantitative data:

Histogram

• Bins or intervals group data (for example, ages 10-19, 20-29, and so on)
• Bars are adjacent to represent continuous intervals on the x axis

Python Example: Histogram

import matplotlib.pyplot as plt

data = [22, 55, 62, 45, 21, 22, 34, 42, 42, 4, 99, 102, 110, 120, 121, 122, 130, 111, 115, 112, 80, 75, 65, 54, 44, 43]
bins = [0, 20, 40, 60, 80, 100, 120, 140]

plt.hist(data, bins=bins, color='orange', edgecolor='black')
plt.xlabel('Value')
plt.ylabel('Frequency')
plt.title('Histogram Example')
plt.show()

---

🧠 3. Types of Histogram Shapes

Histograms don’t just show data — their shapes tell a story.

📈 Bell-Shaped (Normal)

• Most data is in the center
• Few values at the extremes
• Example: IQ scores, height

🔄 Skewed Right (Positive Skew)

• Long tail on the right
• Most values are low
• Example: Income (most people earn little, few earn a lot)

🔃 Skewed Left (Negative Skew)

• Long tail on the left
• Most values are high
• Example: Age of retirement (most people retire around 60–65)

⛰️ Bimodal (Two Peaks)

• Two distinct groups in the data
• Example: Test scores from two different classes

---

🖼️ Visual Guide to Histogram Shapes

---

🤖 Why Data Visualization Matters in Machine Learning

Data visualization is a crucial step in any machine learning workflow:

• Explore distributions: Histograms and bar charts help you spot skewed data, outliers, or class imbalance before modeling.
• Feature selection: Visualizations reveal which variables may be most informative for your model.
• Model diagnostics: After training, graphs help communicate results, feature importance, and errors.

For example, a histogram of your target variable can reveal if you have enough positive and negative cases for a classification task.

---

🔍 ML Tip: Use bar charts to compare the number of samples in each class label (like 0s and 1s in classification problems).

---

✅ Best practices for choosing and designing graphs

• Match the graph to the data type. Use bar charts or pie charts for categorical data and histograms for quantitative data.
• Use high contrast and clear labels. Choose colors with good contrast and always label axes, categories, and titles clearly.
• Sort categories in a meaningful order. For bar charts, sort categories alphabetically or by frequency to make patterns easier to see.
• Show proportions when the total matters. Use pie charts or relative frequencies when you care about parts of a whole, and keep the number of slices small.
• Combine numbers and visuals. Use graphs together with simple tables or summary statistics to tell a complete story about the data.

---

⚠️ Common pitfalls with graphs

• Using pie charts with too many categories. Too many slices make it hard to compare sizes; use a bar chart instead when there are many categories.
• Mixing up bar charts and histograms. Bar charts are for separate categories with gaps between bars, histograms are for continuous intervals and use adjacent bars.
• Manipulating axes in a misleading way. Starting the y axis at a high value or using inconsistent scales can exaggerate or hide differences.
• Ignoring missing or zero values. If you drop missing or zero categories from the graph, the picture of the data can become misleading.
• Overloading graphs with decorations. Heavy gridlines, 3D effects, or too many colors can distract from the actual message of the data.

---

🧠 Level Up: Why Choosing the Right Graph Matters in Data Science

Effective data visualization is more than just making charts look nice — it’s about choosing the right tool to reveal insights clearly and accurately:

• 📊 Bar charts and pie charts work best for categorical data, helping compare groups or parts of a whole.
• 📈 Histograms are ideal for quantitative data, showing distribution shapes like normal, skewed, or bimodal.
• 🔍 The shape of a histogram can hint at underlying processes, identify outliers, and guide statistical modeling decisions.
• 🎯 Choosing the wrong graph can mislead viewers or hide important patterns — so the choice of graph is a vital skill.

Mastering graph selection will make your analyses clearer and your communication more impactful.

---

Audience & Purpose Tip:
Choose your graph based on your audience and your goal. For non-technical readers, simple bar or pie charts work best. For technical analysis, histograms or scatter plots might be more appropriate.

---

Try it yourself:
Use free tools like Plotly, Tableau Public, or Google Sheets to create your own interactive graphs and experiment with different data types!

---

📌 Try It Yourself

Q: You surveyed 200 people to find out their most-used social media platform.

Which type of graph would best show the results?

💡 Show Answer

✅ A bar chart or pie chart — because you're visualizing a categorical (nominal) variable.

Each bar or slice represents the number (or percentage) of people who prefer a platform like Instagram, TikTok, or X (Twitter).

---

Bonus: What if you had their daily screen time in hours instead?

💡 Show Answer

✅ Use a histogram — because screen time is quantitative and continuous.

A histogram shows how the data is distributed across intervals (e.g., 0–2, 2–4, 4–6 hours), which helps identify patterns like clustering or skewness.

---

💬 Got a question or suggestion?
Feel free to leave a comment in the section below — I’d love to hear your thoughts or help with your dataset!

---

🧾 Summary Table

Data Type	Graph Type	Use When…
Nominal	Bar, Pie	Categories with no order
Ordinal	Bar	Ordered categories
Interval/Ratio	Histogram	Numeric data (continuous)

---

✅ Up Next

We’ll build on this and create frequency tables — the building blocks behind many of these graphs!