Why Does y = home_data.SalePrice Work, but y = home_data.[features] Does Not?

[Rajeev Bagra]

When learning Pandas and Machine Learning, many beginners notice that the following code works:

y = home_data.SalePrice

but this does not:

y = home_data.[features]

Why is that?

The Big Idea

Pandas provides two different ways to access columns:

1. Dot notation for a single column
2. Bracket notation for one or more columns

Understanding the difference helps explain why one statement works and the other produces an error.

---

Single Column Access

Suppose our CSV file contains:

LotArea,YearBuilt,SalePrice
8450,2003,208500
9600,1976,181500
11250,2001,223500

After loading the file:

home_data = pd.read_csv(iowa_file_path)

we can access the SalePrice column in two equivalent ways:

y = home_data.SalePrice

or

y = home_data['SalePrice']

Both return the same column.

The dot notation is simply a convenient shortcut when accessing a single column.

---

Multiple Column Access

Now suppose we want several feature columns:

features = ['LotArea', 'YearBuilt', '1stFlrSF']

To select all of them, we use:

X = home_data[features]

which is equivalent to:

X = home_data[['LotArea', 'YearBuilt', '1stFlrSF']]

Here, features is a list containing multiple column names.

---

Why Doesn’t home_data.features Work?

Consider:

features = ['LotArea', 'YearBuilt']
X = home_data.features

Pandas interprets this as:

“Find a column literally named features.”

Since no such column exists, it will fail.

The variable features is a Python list, not a column name.

---

Why Is home_data.[features] Invalid?

This statement:

y = home_data.[features]

is not valid Python syntax.

After a dot (.), Python expects an attribute name:

home_data.SalePrice

but not:

home_data.[features]

The square brackets must be applied directly to the DataFrame:

home_data[features]

---

A Helpful Analogy

Think of a DataFrame as a filing cabinet.

• home_data.SalePrice means: “Give me the drawer named SalePrice.”
• home_data['SalePrice'] means the same thing.
• home_data[features] means: “Give me all drawers listed in the features list.”

Python knows how to fetch one drawer using dot notation, but multiple drawers require bracket notation.

---

Machine Learning Convention

In many machine learning projects, you’ll see:

y = home_data['SalePrice']

features = ['LotArea', 'YearBuilt', '1stFlrSF']
X = home_data[features]

Here:

• y = target column (what we want to predict)
• X = feature columns (information used for prediction)

For house-price prediction:

Features (X)
LotArea
YearBuilt
1stFlrSF
      ↓
Machine Learning Model
      ↓
SalePrice (y)

---

Rule of Thumb

One column:

df.column_name

or

df['column_name']

Multiple columns:

df[list_of_columns]

Remember:

home_data.SalePrice      ✓
home_data['SalePrice']   ✓

home_data[features]      ✓

home_data.features       ✗ (looks for a column named "features")
home_data.[features]     ✗ (invalid syntax)

Once you understand this distinction, many Pandas and Machine Learning examples become much easier to read.

[Author]

Posts