Rolling Window Calculation in Pandas: rolling vs. Manual

Pandas: Performance Optimization Exercise-18 with Solution

Write a Pandas program to perform a rolling window calculation on a time series DataFrame using the rolling method. Compare the performance with manual calculation.

Sample Solution :

Python Code :

# Import necessary libraries
import pandas as pd
import numpy as np
import time

# Create a time series DataFrame
num_rows = 1000000
date_range = pd.date_range(start='1/1/2020', periods=num_rows, freq='T')
df = pd.DataFrame({'value': np.random.randn(num_rows)}, index=date_range)

# Define the window size
window_size = 60

# Measure time for rolling method
start_time = time.time()
rolling_mean = df['value'].rolling(window=window_size).mean()
end_time = time.time()
rolling_time = end_time - start_time

# Measure time for manual rolling calculation
start_time = time.time()
manual_rolling_mean = df['value'].copy()
for i in range(window_size, num_rows):
    manual_rolling_mean.iloc[i] = df['value'].iloc[i-window_size:i].mean()
end_time = time.time()
manual_rolling_time = end_time - start_time

# Print the time taken for each method
print(f"Time taken using rolling method: {rolling_time:.6f} seconds")
print(f"Time taken using manual calculation: {manual_rolling_time:.6f} seconds")

Output:

Time taken using rolling method: 0.034991 seconds
Time taken using manual calculation: 95.308910 seconds

Explanation:

• Import Libraries:
• Import pandas, numpy, and time.
• Create Time Series DataFrame:
• Generate a time series DataFrame with 1,000,000 rows, each representing a minute.
• Define Window Size:
• Set the window size for the rolling calculation (e.g., 60).
• Time Measurement for rolling Method:
• Measure the time taken to calculate the rolling mean using the rolling method.
• Time Measurement for Manual Calculation:
• Measure the time taken to manually calculate the rolling mean using a for loop.
• Print Results:
• Print the time taken for each method.

Python-Pandas Code Editor:

Have another way to solve this solution? Contribute your code (and comments) through Disqus.

Previous: Performance comparison of DataFrame sorting in Pandas.
Next: Efficiently apply multiple Aggregation functions in Pandas.