Python for Data course menu

pandas Data Types: astype, to_numeric, to_datetime and category

Fix column types in pandas: check dtypes, convert with astype, clean numbers stored as text with to_numeric, parse dates with to_datetime, use the category type to save memory, and nullable integer and string types.

Intermediate9 min readUpdated October 24, 2026

On this page

Checking types
astype for clean values
to_numeric for messy numbers
to_datetime for dates
The category type
Nullable types: Int64 and string
Putting it together
Common mistakes
Practice
Key takeaway

Every column has a dtype. When a column has the wrong type, calculations fail or give nonsense: numbers stored as text sort alphabetically and cannot be summed, and dates stored as text cannot be compared by month. Fixing types is a core cleaning step.

Checking types#

Python

import pandas as pd

raw = pd.DataFrame({
    "order_id": ["001", "002", "003", "004"],
    "price":    ["1,200", "850", "2,499", "n/a"],
    "quantity": ["2", "1", "3", "1"],
    "ordered":  ["05/01/2026", "09/01/2026", "11/02/2026", "20/02/2026"],
    "status":   ["paid", "paid", "pending", "paid"],
})
print(raw.dtypes)
print(raw["quantity"].sum())      # text joined together, not added!

Output

order_id    object
price       object
quantity    object
ordered     object
status      object
dtype: object
2131

Everything was read as text (object), so "summing" the quantities just joins the strings.

astype for clean values#

When every value converts cleanly, astype does the job:

Python

import pandas as pd

quantity = pd.Series(["2", "1", "3", "1"])
q = quantity.astype(int)
print(q.dtype, q.sum())
print(pd.Series([1.0, 2.0]).astype(int).tolist(), pd.Series([1, 0, 1]).astype(bool).tolist())

Output

int64 7
[1, 2] [True, False, True]

to_numeric for messy numbers#

astype(float) fails on values like "1,200" or "n/a". Clean the text first, then use pd.to_numeric with errors="coerce", which turns anything still invalid into NaN:

Python

import pandas as pd

price = pd.Series(["1,200", "850", "2,499", "n/a"])
try:
    price.astype(float)
except ValueError as e:
    print("Error:", e)

clean = pd.to_numeric(price.str.replace(",", ""), errors="coerce")
print(clean)

Output

Error: could not convert string to float: '1,200'
0    1200.0
1     850.0
2    2499.0
3       NaN
dtype: float64

After coercing, check how many values became NaN (clean.isna().sum()): each one is a value you could not read.

to_datetime for dates#

pd.to_datetime turns text into real timestamps. Give it the format so that day and month are never confused (05/01 could be 5 January or 1 May):

Python

import pandas as pd

ordered = pd.Series(["05/01/2026", "09/01/2026", "11/02/2026", "20/02/2026"])
dates = pd.to_datetime(ordered, format="%d/%m/%Y")
print(dates)
print(dates.dt.month.tolist(), dates.dt.day_name().tolist())
print((dates.max() - dates.min()).days, "days between first and last")

Output

0   2026-01-05
1   2026-01-09
2   2026-02-11
3   2026-02-20
dtype: datetime64[ns]
[1, 1, 2, 2] ['Monday', 'Friday', 'Wednesday', 'Friday']
46 days between first and last

Common format codes: %d day, %m month, %Y four-digit year, %y two-digit year, %H:%M hours and minutes, %b month name like Jan. More in Dates and Time Series.

The category type#

A text column with few distinct values (statuses, cities, sizes) can be stored as category: each value is stored once and rows hold small codes. It uses less memory and can have a meaningful order:

Python

import pandas as pd

status = pd.Series(["paid", "pending", "paid", "refunded", "paid"] * 20000)
as_category = status.astype("category")
print(status.memory_usage(deep=True), as_category.memory_usage(deep=True))
print(as_category.cat.categories.tolist())

sizes = pd.Series(["M", "S", "L", "M"]).astype(pd.CategoricalDtype(["S", "M", "L"], ordered=True))
print(sizes.sort_values().tolist(), (sizes > "S").tolist())

Output

5440132 100406
['paid', 'pending', 'refunded']
['S', 'M', 'M', 'L'] [True, False, True, True]

The category column uses far less memory. With an ordered category, sorting and comparisons follow S < M < L instead of alphabetical order.

Nullable types: Int64 and string#

A regular integer column cannot hold NaN (NaN is a float), so one missing value turns a whole column into floats. pandas' nullable types fix that, using <NA> for missing values:

Python

import pandas as pd

ages = pd.Series([25, None, 31])
print(ages.dtype, ages.tolist())

ages_int = pd.Series([25, None, 31], dtype="Int64")      # capital I
print(ages_int.dtype, ages_int.tolist())

names = pd.Series(["Asha", None], dtype="string")
print(names.dtype, names.str.upper().tolist())

Output

float64 [25.0, nan, 31.0]
Int64 [25, <NA>, 31]
string ['ASHA', <NA>]

convert_dtypes() converts a whole DataFrame to these nullable types in one step.

Putting it together#

Python

import pandas as pd

raw = pd.DataFrame({
    "order_id": ["001", "002", "003", "004"],
    "price":    ["1,200", "850", "2,499", "n/a"],
    "quantity": ["2", "1", "3", "1"],
    "ordered":  ["05/01/2026", "09/01/2026", "11/02/2026", "20/02/2026"],
    "status":   ["paid", "paid", "pending", "paid"],
})
clean = raw.assign(
    price=pd.to_numeric(raw["price"].str.replace(",", ""), errors="coerce"),
    quantity=raw["quantity"].astype(int),
    ordered=pd.to_datetime(raw["ordered"], format="%d/%m/%Y"),
    status=raw["status"].astype("category"),
)
print(clean.dtypes)
print((clean["price"] * clean["quantity"]).sum())

Output

order_id            object
price              float64
quantity             int64
ordered     datetime64[ns]
status            category
dtype: object
10747.0

order_id stays as text on purpose: it is a label, not a number, and its leading zeros matter.

Common mistakes#

Practice#

Practice exercise

Clean pd.Series(["₹1,250", "₹980", "free", "₹2,100"]) into numbers, treating "free" as 0. Print the cleaned values as integers and their total.

Show hint

Remove ₹ and , with str.replace, use to_numeric(errors="coerce"), then fillna(0) and astype(int).

Show solution

Python

import pandas as pd

prices = pd.Series(["₹1,250", "₹980", "free", "₹2,100"])
numbers = pd.to_numeric(prices.str.replace("₹", "").str.replace(",", ""), errors="coerce")
numbers = numbers.fillna(0).astype(int)
print(numbers.tolist(), numbers.sum())

Output

[1250, 980, 0, 2100] 4330

Key takeaway#

Next: Cleaning Text Columns.

Finished this lesson?

Python for DataHandling Missing Values in pandas: isna, dropna and fillnaFind, count, drop and fill missing data in pandas: NaN and None, isna and notna, dropna with thresholds, fillna with constants, means and group means, forward fill and interpolation, and how missing values affect calculations.Intermediate12 min read Python for DataCleaning Text Data in pandas with .str MethodsClean messy text columns in pandas: strip spaces, change case, replace characters, split columns, extract values with regular expressions, check text with contains and startswith, and standardise inconsistent labels.Intermediate7 min read Python for DataDates and Time Series in pandas: to_datetime, dt and resampleWork with dates in pandas: parse dates, extract year, month and weekday with .dt, date arithmetic with Timedelta, date ranges, filtering by date, a DatetimeIndex, and resampling daily data to weekly or monthly totals.Intermediate9 min read

Checking types#

astype for clean values#

to_numeric for messy numbers#

to_datetime for dates#

The category type#

Nullable types: Int64 and string#

Putting it together#

Common mistakes#

Practice#

Key takeaway#

Related lessons