Accessing DataFrame Columns in Pandas: `df.column` vs `df[“column”]`

[Rajeev Bagra]

When working with Pandas DataFrames, one common question beginners ask is:

Should I access a column using df.price_usd or df["price_usd"]?

Both work — but they are not equally reliable.

Let’s break it down clearly.

---

ߧ The Scenario

Suppose you’re plotting a histogram:

fig, ax = plt.subplots()
ax.hist(df["price_usd"][:20000])

You might wonder:

Could I just use df.price_usd instead?

The answer is yes — but with conditions.

---

✅ Method 1: Dot Notation (df.price_usd)

ax.hist(df.price_usd[:20000])

✔️ When It Works

• Column name has no spaces
• No special characters
• Does not conflict with existing DataFrame methods

Example of a valid column:

price_usd

---

❌ When It Breaks

1. Column with spaces:

df.home price # ❌ Error

1. Column conflicting with built-in methods:

df.mean # ❌ Refers to method, not column

So dot notation is convenient — but fragile.

---

✅ Method 2: Bracket Notation (df["price_usd"]) — Recommended

ax.hist(df["price_usd"][:20000])

✔️ Why This Is Better

• Works with any column name
• Handles spaces and special characters
• Avoids conflicts with DataFrame methods
• Standard in professional and production code

Even this works perfectly:

df["home price"]
df["mean"]
df["price-usd"]

No surprises. No hidden errors.

---

ߎ When Should You Create a Separate Variable?

Instead of writing:

ax.hist(df["price_usd"][:20000])

You might sometimes see:

price_usd = df["price_usd"]
ax.hist(price_usd[:20000])

✔️ This is useful when:

• You reuse the column multiple times
• You calculate statistics (mean, median, etc.)
• You want cleaner, more readable code

Example:

price_usd = df["price_usd"]

mean_price = price_usd.mean()
median_price = price_usd.median()
ax.hist(price_usd[:20000])

That improves clarity.

---

ߓ Best Practice Summary

Method	Works Always?	Professional Code?	Recommended?
df.price_usd	❌ No	Rarely	⚠️ Use Carefully
df["price_usd"]	✅ Yes	Yes	✅ Recommended

---

ߚ Final Takeaway

For learning:

• Dot notation is quick and readable.
• Bracket notation is robust and production-safe.

If you’re building serious data projects, always prefer:

df["column_name"]

It saves you from subtle bugs later.

---

[Author]

Posts