Using appropriate dtypes in Pandas - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of memory usage with different dtypes

What is the output of the following code snippet that compares memory usage of integer columns with default and optimized dtypes?

Pandas

import pandas as pd
import numpy as np

df = pd.DataFrame({
    'A': np.random.randint(0, 100, size=1000),
    'B': np.random.randint(0, 100, size=1000)
})

mem_default = df.memory_usage(deep=True, index=False).sum()
df['A'] = df['A'].astype('int8')
df['B'] = df['B'].astype('int8')
mem_optimized = df.memory_usage(deep=True, index=False).sum()
print(mem_default, mem_optimized)

A 16000 2000

B 16000 4000

C 8000 4000

D 8000 2000

💡 Hint

Think about how changing from default int64 to int8 affects memory size.

The default integer dtype in pandas is int64, which uses 8 bytes per value. Changing to int8 reduces each value to 1 byte. For 1000 rows and 2 columns, default memory is 8*1000*2=16000 bytes. After conversion, 1*1000*2=2000 bytes. So output is '16000 2000'.

❓ data_output

intermediate

Resulting dtypes after conversion

Given the DataFrame below, what are the dtypes of columns after applying the conversion code?

Pandas

import pandas as pd

df = pd.DataFrame({
    'col1': [1, 2, 3],
    'col2': [0.1, 0.2, 0.3],
    'col3': ['a', 'b', 'c']
})

df['col1'] = df['col1'].astype('int16')
df['col2'] = df['col2'].astype('float32')

A col1: int16, col2: float32, col3: object

B col1: int64, col2: float64, col3: object

C col1: int16, col2: float64, col3: string

D col1: int32, col2: float32, col3: object

💡 Hint

Check the astype conversions applied to col1 and col2.

The code explicitly converts col1 to int16 and col2 to float32. col3 remains unchanged as object dtype because it contains strings.

🔧 Debug

advanced

Identify the error in dtype conversion

What error will this code raise when trying to convert the 'age' column to 'int8'?

Pandas

import pandas as pd

df = pd.DataFrame({'age': [25, 300, 45]})
df['age'] = df['age'].astype('int8')

A TypeError

B ValueError

C OverflowError

D No error, conversion succeeds

💡 Hint

Check if values fit in int8 range (-128 to 127).

Although 300 is outside int8 range, pandas silently wraps around values when converting to int8, so no error is raised. The value 300 becomes -56 due to overflow.

🧠 Conceptual

advanced

Best dtype for categorical data

Which dtype is most memory efficient and appropriate for a column with repeated string categories like 'red', 'blue', 'green'?

A object

B category

C string

D int64

💡 Hint

Think about how pandas stores repeated categories internally.

The 'category' dtype stores repeated string values as integer codes internally, saving memory and improving performance for repeated categories.

🚀 Application

expert

Optimize memory usage for mixed dtype DataFrame

Given a DataFrame with columns: 'id' (integers 0-100000), 'score' (floats 0-1), 'grade' (strings 'A', 'B', 'C'), which dtype conversions will minimize memory usage without losing data?

A id: int8, score: float32, grade: category

B id: int16, score: float64, grade: object

C id: int32, score: float32, grade: category

D id: int64, score: float16, grade: string

💡 Hint

Consider the range of 'id' and precision needed for 'score'.

'id' values up to 100000 fit in int32 but not int16 or int8. 'score' between 0 and 1 can use float32 safely. 'grade' with few repeated strings is best as category dtype.