如何在Python3中将指定NumPy数组写入带表头的CSV文件

Hey there! Let's break down how to get those two NumPy arrays into a CSV file with the headers "date" and "value". Since you're already working with a DataFrame (covid_df), the simplest approach uses pandas (which you're likely already using), but I'll also cover a couple of alternatives for different use cases.

Method 1: Use Pandas (Recommended) #

This is the most straightforward option, especially since you already have a DataFrame. Even if you've converted columns to NumPy arrays, you can easily reconstruct a new DataFrame and save it directly:

import pandas as pd
import numpy as np

# Your existing NumPy arrays
date_array = np.array(covid_df['datetime'])
value_array = np.array(covid_df['value'])

# Create a new DataFrame with the desired column names
output_df = pd.DataFrame({
    'date': date_array,
    'value': value_array
})

# Write to CSV – index=False prevents an extra index column from being added
output_df.to_csv('covid_data.csv', index=False)

Why this works: #

Pandas handles all the heavy lifting for CSV formatting, including writing the headers correctly and ensuring data types are preserved. The index=False parameter is crucial here—it stops pandas from adding an unnecessary index column to your output file.

Method 2: Pure NumPy with savetxt #

If you want to stick to just NumPy without pandas, you can use np.savetxt by combining your arrays first:

import numpy as np

date_array = np.array(covid_df['datetime'])
value_array = np.array(covid_df['value'])

# Combine the two arrays side-by-side (column-wise)
combined_data = np.column_stack((date_array, value_array))

# Write to CSV – note the comments='' to avoid a # prefix on the header
np.savetxt(
    'covid_data.csv',
    combined_data,
    delimiter=',',
    header='date,value',
    fmt='%s',  # Use %s to handle string dates; adjust for value if needed (e.g., %f for floats)
    comments=''
)

Notes: #

• The fmt parameter defines how each column is formatted. If your value array contains numbers, you can specify separate formats like fmt=['%s', '%f'] to handle dates as strings and values as floats.
• comments='' removes the default # that NumPy adds to the header line.

Method 3: Python's Built-in csv Module #

For a more low-level approach, you can use Python's native csv module to write rows manually:

import csv
import numpy as np

date_array = np.array(covid_df['datetime'])
value_array = np.array(covid_df['value'])

# Open the CSV file in write mode
with open('covid_data.csv', 'w', newline='') as csv_file:
    writer = csv.writer(csv_file)
    # Write the header row first
    writer.writerow(['date', 'value'])
    # Iterate over paired elements from both arrays and write each row
    for date, val in zip(date_array, value_array):
        writer.writerow([date, val])

Why this works: #

The zip function pairs elements from date_array and value_array one-by-one, and csv.writer handles formatting each row correctly. The newline='' parameter prevents extra blank lines from appearing in the output file on Windows systems.

内容的提问来源于stack exchange，提问作者user3191435