Pandas and NumPy connection - Time & Space Complexity

Time Complexity: Pandas and NumPy connection

O(n)

Understanding Time Complexity

We want to see how fast pandas works when it uses NumPy arrays inside. This helps us know how the time to run grows when data gets bigger.

How does the time to do operations change as the data size grows?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

import pandas as pd
import numpy as np

arr = np.arange(1000)
df = pd.DataFrame({'numbers': arr})
df['squared'] = df['numbers'] ** 2

This code creates a pandas DataFrame from a NumPy array and adds a new column by squaring the numbers.

Identify Repeating Operations

• Primary operation: Squaring each number in the 'numbers' column.
• How many times: Once for each element in the array (n times).

How Execution Grows With Input

As the number of rows grows, the time to square each number grows too, because each number needs to be processed.

Input Size (n)	Approx. Operations
10	10 squaring operations
100	100 squaring operations
1000	1000 squaring operations

Pattern observation: The operations grow directly with the number of items; doubling the data doubles the work.

Final Time Complexity

Time Complexity: O(n)

This means the time to run grows in a straight line with the number of rows in the DataFrame.

Common Mistake

[X] Wrong: "Using pandas with NumPy arrays makes operations instant, no matter the size."

[OK] Correct: Even though pandas uses fast NumPy arrays, it still needs to do work for each item, so time grows with data size.

Interview Connect

Understanding how pandas and NumPy work together helps you explain data processing speed clearly. This skill shows you know what happens behind the scenes when working with data.

Self-Check

"What if we used a vectorized NumPy function directly on the array instead of pandas? How would the time complexity change?"

Practice

1. What is the main reason Pandas uses NumPy arrays internally?

easy

A. To handle text data only

B. To make data visualization easier

C. To store data efficiently and allow fast computations

D. To create interactive dashboards

Solution

1. Step 1: Understand Pandas data storage

   Pandas stores data in tables called DataFrames, which need efficient storage for speed.

2. Step 2: Role of NumPy arrays

   NumPy arrays provide fast, memory-efficient storage and operations on numerical data.

3. Final Answer:

   To store data efficiently and allow fast computations -> Option C

4. Quick Check:

   Pandas uses NumPy arrays for efficiency [OK]

Hint: Pandas uses NumPy arrays for speed and memory [OK]

Common Mistakes:

• Thinking Pandas uses NumPy only for visualization
• Believing Pandas stores data as plain Python lists
• Confusing NumPy arrays with text handling

2. Which of the following is the recommended way to convert a Pandas DataFrame df to a NumPy array?

easy

A. np.array(df)

B. df.to_numpy()

C. df.numpy()

D. df.asarray()

Solution

1. Step 1: Recall Pandas method for conversion

   Pandas DataFrames have a method to_numpy() to convert to NumPy arrays.

2. Step 2: Check other options

   np.array(df) works but is not the dedicated Pandas method; df.numpy() and df.asarray() do not exist.

3. Final Answer:

   df.to_numpy() -> Option B

4. Quick Check:

   Use to_numpy() to convert DataFrame [OK]

Hint: Use df.to_numpy() to get NumPy array from DataFrame [OK]

Common Mistakes:

• Using non-existent methods like df.numpy()
• Confusing np.array(df) as the best method
• Trying to use df.asarray() which is invalid

3. Given the code:

import pandas as pd
import numpy as np
df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
result = np.sum(df)

What is the value of result?

medium

A. array([6, 15])

B. [6 15]

C. 21

D. Error: cannot sum DataFrame with NumPy

Solution

1. Step 1: Understand np.sum on DataFrame

   Applying np.sum to a DataFrame sums each column and returns a NumPy array.

2. Step 2: Calculate sums of columns

   Column 'A' sums to 1+2+3=6, column 'B' sums to 4+5+6=15, so result is array([6, 15]).

3. Final Answer:

   array([6, 15]) -> Option A

4. Quick Check:

   NumPy sums columns into array [OK]

Hint: np.sum on DataFrame sums columns, returns NumPy array [OK]

Common Mistakes:

• Expecting a single number sum of all values
• Thinking np.sum causes error on DataFrame
• Confusing list output with NumPy array

4. What is wrong with this code snippet?

import pandas as pd
import numpy as np
df = pd.DataFrame({'X': [1, 2], 'Y': [3, 4]})
arr = df.values()
print(arr + 1)

medium

A. df.values() should be df.values without parentheses

B. You cannot add 1 to a NumPy array

C. DataFrame does not have a values attribute

D. The print statement syntax is incorrect

Solution

1. Step 1: Check usage of df.values

   df.values is an attribute, not a method, so parentheses cause an error.

2. Step 2: Understand addition to NumPy array

   Adding 1 to a NumPy array works element-wise, so no error there.

3. Final Answer:

   df.values() should be df.values without parentheses -> Option A

4. Quick Check:

   Use df.values without () to get NumPy array [OK]

Hint: df.values is attribute, not method; no parentheses [OK]

Common Mistakes:

• Calling df.values() as a method
• Thinking addition to array is invalid
• Assuming print syntax is wrong in Python 3

5. You have a Pandas DataFrame df with numeric columns. You want to apply the NumPy function np.log to all values and get a new DataFrame with the results preserving the original index and column names. Which code correctly does this?

hard

A. new_df = pd.DataFrame(np.log(df))

B. new_df = df.to_numpy().log()

C. new_df = np.log(df)

D. new_df = df.apply(np.log)

Solution

1. Step 1: Understand applying NumPy functions on DataFrame

   Using df.apply(np.log) applies np.log element-wise to each column and returns a DataFrame preserving index and columns.

2. Step 2: Check other options

   np.log(df) returns a NumPy array, not DataFrame; pd.DataFrame(np.log(df)) creates DataFrame but loses original index and column names; df.to_numpy().log() is invalid syntax.

3. Final Answer:

   new_df = df.apply(np.log) -> Option D

4. Quick Check:

   Use df.apply(np.log) for DataFrame output [OK]

Hint: Use df.apply(np.log) to keep DataFrame structure [OK]

Common Mistakes:

• Using np.log(df) which returns array, not DataFrame
• Trying df.to_numpy().log() which is invalid
• Wrapping np.log(df) inside pd.DataFrame unnecessarily