Resampling with groupby for time data in Pandas - Time & Space Complexity

Time Complexity: Resampling with groupby for time data

O(g * n)

Understanding Time Complexity

We want to understand how the time to resample grouped time data changes as the data grows.

How does the work increase when we have more groups or more time points?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

import pandas as pd

df = pd.DataFrame({
    'date': pd.date_range('2023-01-01', periods=1000, freq='H'),
    'group': ['A', 'B'] * 500,
    'value': range(1000)
})

result = df.groupby('group').resample('D', on='date').sum()

This code groups data by a category, then resamples each group by day, summing values.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: Grouping the data and then resampling each group separately.
• How many times: The resampling runs once per group, and inside each group it processes all time points.

How Execution Grows With Input

As the number of groups or time points grows, the work grows roughly by multiplying these two.

Input Size (n)	Approx. Operations
10 groups, 100 time points each	~1,000 operations
100 groups, 1,000 time points each	~100,000 operations
1,000 groups, 10,000 time points each	~10,000,000 operations

Pattern observation: The total work grows roughly by multiplying the number of groups and the number of time points per group.

Final Time Complexity

Time Complexity: O(g * n)

This means the time grows proportionally to the number of groups times the number of time points in each group.

Common Mistake

[X] Wrong: "Resampling once on the whole data is the same speed as resampling after grouping."

[OK] Correct: Grouping splits data, so resampling runs separately on each group, multiplying the work by the number of groups.

Interview Connect

Understanding how grouping and resampling scale helps you handle real-world time data efficiently and explain your approach clearly.

Self-Check

"What if we resampled before grouping instead of after? How would the time complexity change?"