Common dtype errors and fixes in Pandas - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output of this dtype conversion code?

Consider the following pandas code that attempts to convert a column to numeric. What will be the output of df['A'] after running this code?

Pandas

import pandas as pd

df = pd.DataFrame({'A': ['1', '2', 'three', '4']})
df['A'] = pd.to_numeric(df['A'], errors='coerce')
print(df['A'])

A Raises ValueError due to 'three' string

B

0       1
1       2
2    three
3       4
Name: A, dtype: object

C

0    1.0
1    2.0
2    NaN
3    4.0
Name: A, dtype: float64

D

0    1
1    2
2    0
3    4
Name: A, dtype: int64

💡 Hint

Look at how errors='coerce' handles non-numeric values.

The pd.to_numeric function converts values to numbers. When errors='coerce' is set, any value that cannot be converted becomes NaN. So 'three' becomes NaN, and the column dtype becomes float64 because NaN is a float.

❓ data_output

intermediate

How many unique values are in the column after conversion?

Given this code, how many unique values does the column 'B' have after conversion?

Pandas

import pandas as pd

df = pd.DataFrame({'B': ['10', '20', '30', '20', 'ten']})
df['B'] = pd.to_numeric(df['B'], errors='coerce')
unique_count = df['B'].nunique(dropna=True)
print(unique_count)

A 3

B 4

C 5

D Raises TypeError

💡 Hint

Remember that nunique(dropna=True) excludes NaN values.

The values after conversion are 10.0, 20.0, 30.0, 20.0, and NaN (from 'ten'). The unique non-NaN values are 10.0, 20.0, and 30.0, so the count is 3.

🔧 Debug

advanced

Why does this code raise a ValueError?

This code tries to convert a column to datetime but raises a ValueError. What is the cause?

Pandas

import pandas as pd

df = pd.DataFrame({'dates': ['2023-01-01', 'not_a_date', '2023-03-01']})
df['dates'] = pd.to_datetime(df['dates'], format='%Y-%m-%d')

A Because the format string is incorrect for the given dates

B Because 'not_a_date' does not match the format and no error handling is set

C Because pd.to_datetime cannot convert strings to datetime

D Because the DataFrame column is not of string type

💡 Hint

Check how pd.to_datetime behaves when a string does not match the format and no errors parameter is given.

The format parameter enforces strict parsing. The string 'not_a_date' does not match the format '%Y-%m-%d', so pd.to_datetime raises a ValueError unless errors='coerce' or errors='ignore' is set.

🚀 Application

advanced

Which option correctly converts a mixed-type column to numeric with invalid values as zero?

You have a pandas column with numbers as strings and some invalid strings. You want to convert it to numeric, replacing invalid values with zero. Which code does this correctly?

Pandas

import pandas as pd

df = pd.DataFrame({'vals': ['5', '10', 'abc', '20']})

A df['vals'] = pd.to_numeric(df['vals'], errors='raise')

B df['vals'] = pd.to_numeric(df['vals'], errors='ignore').fillna(0)

C df['vals'] = df['vals'].astype(int)

D df['vals'] = pd.to_numeric(df['vals'], errors='coerce').fillna(0).astype(int)

💡 Hint

Think about how to handle invalid strings and convert the column to integers.

Option D converts invalid strings to NaN with errors='coerce', then replaces NaN with 0, and finally converts the column to integers.

Option D does not convert invalid strings and fillna(0) will not work properly. Option D raises an error because 'abc' cannot be cast to int. Option D raises an error on invalid strings.

🧠 Conceptual

expert

What is the main reason pandas converts integer columns with NaN to float dtype?

In pandas, when you convert a column with integers but some values are missing (NaN), the column dtype changes from int to float. Why does this happen?

A Because NaN is a float value and integer dtype cannot represent NaN

B Because pandas automatically converts all columns with missing values to float for performance

C Because integers are stored as floats internally in pandas

D Because NaN is treated as zero in integer columns

💡 Hint

Think about how missing values are represented in pandas and numpy.

NaN is a special floating-point value. Integer types in pandas (and numpy) cannot represent NaN. So when a column has integers and NaN, pandas upcasts the column to float to accommodate NaN.