Resampling time series in Data Analysis Python

Introduction

Resampling helps change the frequency of time series data. It lets you see data in bigger or smaller time steps.

You have daily sales data but want to see monthly totals.

You want to convert minute-level temperature readings to hourly averages.

You need to fill missing dates in your data with estimated values.

You want to downsample high-frequency stock prices to daily closing prices.

Syntax

Data Analysis Python

df.resample('rule').aggregation_function()

rule is a string like 'D' for day, 'M' for month, 'H' for hour.

You must use an aggregation function like mean(), sum(), or count() after resampling.

Examples

Sum values for each month.

Data Analysis Python

df.resample('M').sum()

Calculate average for each hour.

Data Analysis Python

df.resample('H').mean()

Find maximum value for each week.

Data Analysis Python

df.resample('W').max()

Sample Program

This code creates daily data from June 1 to June 10 with values 1 to 10. Then it resamples the data by week and sums the values in each week.

Data Analysis Python

import pandas as pd
import numpy as np

# Create sample daily data for 10 days
rng = pd.date_range('2024-06-01', periods=10, freq='D')
data = np.arange(10) + 1  # values 1 to 10
df = pd.DataFrame({'value': data}, index=rng)

# Resample to weekly sum
weekly_sum = df.resample('W').sum()

print('Original daily data:')
print(df)
print('\nWeekly sum data:')
print(weekly_sum)

Important Notes

Make sure your DataFrame index is a datetime type before resampling.

Resampling up (to higher frequency) usually needs a method like ffill() or bfill() to fill new rows.

Summary

Resampling changes the time frequency of your data.

Use a string rule like 'D', 'W', 'M' to set the new frequency.

Always apply an aggregation function after resampling.

Practice

1. What does resampling a time series data mean?

easy

A. Removing missing values

B. Sorting the data by values

C. Changing the time frequency of the data

D. Converting data to strings

Solution

1. Step 1: Understand the meaning of resampling

   Resampling means changing how often data points appear in time, like from daily to monthly.

2. Step 2: Compare options to definition

   Only Changing the time frequency of the data correctly describes changing the time frequency of data.

3. Final Answer:

   Changing the time frequency of the data -> Option C

4. Quick Check:

   Resampling = Changing frequency [OK]

Hint: Resampling changes how often data points appear [OK]

Common Mistakes:

• Thinking resampling sorts data
• Confusing resampling with cleaning data
• Believing resampling changes data type

2. Which of the following is the correct syntax to resample a pandas DataFrame df to monthly frequency and take the mean?

easy

A. df.resample('month').mean()

B. df.resample('monthly').mean()

C. df.resample('M').average()

D. df.resample('M').mean()

Solution

1. Step 1: Identify the correct frequency string

   Pandas uses 'M' for monthly frequency, not 'monthly' or 'month'.

2. Step 2: Check aggregation function name

   The correct function is mean(), not average().

3. Final Answer:

   df.resample('M').mean() -> Option D

4. Quick Check:

   Use 'M' and mean() for monthly average [OK]

Hint: Use 'M' for month and mean() to aggregate [OK]

Common Mistakes:

• Using 'monthly' or 'month' instead of 'M'
• Using average() instead of mean()
• Missing parentheses after mean

3. Given the following code, what will be the output?

import pandas as pd
idx = pd.date_range('2023-01-01', periods=4, freq='D')
data = [10, 20, 30, 40]
df = pd.DataFrame({'value': data}, index=idx)
resampled = df.resample('2D').sum()
print(resampled)

medium

A. 2023-01-01 30
2023-01-03 70

B. 2023-01-01 10
2023-01-03 30
2023-01-05 40

C. 2023-01-01 30
2023-01-03 40

D. 2023-01-01 10
2023-01-02 20
2023-01-03 30
2023-01-04 40

Solution

1. Step 1: Understand the original data and frequency

   Data has values 10,20,30,40 on consecutive days from 2023-01-01 to 2023-01-04.

2. Step 2: Resample with '2D' and sum

   Resampling groups every 2 days: (Jan 1+2) and (Jan 3+4). Sums are 10+20=30 and 30+40=70.

3. Final Answer:

   2023-01-01 30 2023-01-03 70 -> Option A

4. Quick Check:

   Sum every 2 days = 30, 70 [OK]

Hint: Sum values in 2-day groups starting from first date [OK]

Common Mistakes:

• Summing only one day instead of two
• Confusing sum with mean
• Misreading the date index

4. What is wrong with this code snippet?

df = pd.DataFrame({'value': [1, 2, 3]}, index=pd.date_range('2023-01-01', periods=3))
resampled = df.resample('W').mean

medium

A. Missing parentheses after 'mean' function call

B. Wrong frequency string 'W' for weekly resampling

C. DataFrame index is not a datetime index

D. Aggregation function 'mean' does not exist

Solution

1. Step 1: Check the resample syntax

   The code uses mean without parentheses, so it returns a method, not the result.

2. Step 2: Verify frequency and index

   'W' is correct for weekly, and the index is a datetime index, so no issues there.

3. Final Answer:

   Missing parentheses after 'mean' function call -> Option A

4. Quick Check:

   Call mean() with () to get result [OK]

Hint: Add () after mean to execute function [OK]

Common Mistakes:

• Forgetting parentheses on aggregation functions
• Thinking 'W' is invalid frequency
• Not setting datetime index before resampling

5. You have minute-level temperature data in a DataFrame df indexed by datetime. You want to resample it to hourly data but only keep hours where the average temperature is above 20°C. Which code snippet correctly does this?

hard

A. df.resample('H').mean()[lambda x: x['temperature'] > 20]

B. df.resample('H').mean().query('temperature > 20')

C. df.resample('H').mean()[df.resample('H').mean() > 20]

D. df.resample('H').mean().filter(lambda x: x > 20)

Solution

1. Step 1: Resample to hourly and calculate mean

   Use df.resample('H').mean() to get hourly averages.

2. Step 2: Filter rows where temperature > 20

   Using .query('temperature > 20') filters rows correctly by the 'temperature' column.

3. Final Answer:

   df.resample('H').mean().query('temperature > 20') -> Option B

4. Quick Check:

   Use query() to filter by column condition [OK]

Hint: Use query() after resample mean to filter rows [OK]

Common Mistakes:

• Applying boolean filter on DataFrame without column name
• Using filter() incorrectly for row filtering
• Repeating resample twice causing inefficiency