Data types

Data types (dtype) are categorized as numeric, categorical, and text.

Check the types

Numbers read from CSV sometimes arrive as text (object), because of values like "1.200", "?" or "35 kg" containing a unit. dtypes shows this:

print(df.dtypes)
df["fiyat"] = pd.to_numeric(df["fiyat"], errors="coerce")

errors="coerce" turns values that can't be converted into NaN; they are then handled as missing values.

Type conversion and categories

The type is changed with astype. Text columns with few distinct values save memory when converted to the category type:

df["yas"] = df["yas"].astype(int)
df["sehir"] = df["sehir"].astype("category")
print(df["sehir"].cat.categories)

The model still doesn't accept text; categories are encoded as numbers in the next lesson.

Exercise Convert to numbers

Convert the ham list to numbers with pd.to_numeric(..., errors="coerce"); print the number of missing values with isna().sum() and the mean of the valid values with mean().

Show example solution

import pandas as pd

s = pd.Series(["10", "20", "?", "30", "yok"])

x = pd.to_numeric(s, errors="coerce")
print(x.isna().sum())
print(x.mean())

Test yourself

What does <code>pd.to_numeric(["5", "?"], errors="coerce")</code> produce?
- [5, 0]
- ["5", "?"]
- [5.0, NaN]
- An error
Show answer

[5.0, NaN] — The unconvertible "?" becomes NaN; because of NaN the result is float.

Convert text to numbers; let unconvertible ones become NaN.

df["fiyat"] = pd.to_numeric(df["fiyat"], errors=____)

Show answer

df["fiyat"] = pd.to_numeric(df["fiyat"], errors="coerce")

Match the types.

int64

Integer
float64

Decimal number
object

Text
category

Limited set of values

Next lesson: Label encoding

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies