如何为大型DataFrame快速将时间值映射为对应班次（Shift）？

If you're working with a large DataFrame, vectorized operations are key to keeping things fast—avoid loops at all costs! Here's how to convert your time values to the corresponding shift labels efficiently:

Step 1: Ensure Your Time Column is in a Datetime Format #

First, convert your string-based Time column to a pandas datetime object (this makes extracting hours straightforward):

import pandas as pd
import numpy as np

# Sample DataFrame matching your structure
data = {
    'Row': [1,2,3,4,5,6,7],
    'Time': ['01:15:12','09:18:22','21:56:01','13:33:23','11:59:56','17:08:38','21:55:16']
}
df = pd.DataFrame(data)

# Convert Time column to datetime (ignores date, focuses on time)
df['Time'] = pd.to_datetime(df['Time'], format='%H:%M:%S')

Step 2: Map Hours to Shifts Using Vectorized Logic #

We have two great, efficient options here—both avoid slow row-wise operations:

Option 1: Using pd.cut (Clean and Concise) #

This method bins hour values directly into your predefined intervals:

# Define bins (0-8, 8-16, 16-24) and corresponding shift labels
bins = [0, 8, 16, 24]
labels = [1, 2, 3]

# Extract hour from datetime and apply binning
df['Shift'] = pd.cut(df['Time'].dt.hour, bins=bins, labels=labels, include_lowest=True)

• include_lowest=True ensures 00:00:00 is correctly grouped into Shift 1.
• Bins are exclusive of upper bounds, so 08:00:00 falls into Shift 2 and 16:00:00 into Shift 3—perfect for your rules.

Option 2: Using np.select (Flexible for Complex Rules) #

If you need explicit control over conditions (great if rules ever change), use np.select:

# Define conditions for each shift
conditions = [
    df['Time'].dt.hour < 8,
    (df['Time'].dt.hour >= 8) & (df['Time'].dt.hour < 16),
    df['Time'].dt.hour >= 16
]
# Corresponding shift values
choices = [1, 2, 3]

# Apply conditions to create Shift column
df['Shift'] = np.select(conditions, choices)

Step 3: Clean Up (Optional) #

If you don't want the default date (1900-01-01) attached to your Time values, convert it back to a time string:

df['Time'] = df['Time'].dt.time

Final Result #

Your DataFrame will now look exactly like what you requested:

Row      Time Shift
0    1  01:15:12     1
1    2  09:18:22     2
2    3  21:56:01     3
3    4  13:33:23     2
4    5  11:59:56     2
5    6  17:08:38     3
6    7  21:55:16     3

Both methods leverage pandas/numpy's optimized vectorized operations, making them ideal for large datasets—they’ll run orders of magnitude faster than manual row loops.

内容的提问来源于stack exchange，提问作者Hussain Rahiminejad