如何用函数实现Pandas Series月度数据统计与可视化（替代硬编码）

Hey there! I totally get where you're coming from—hardcoding each month is such a tedious, non-scalable hassle. Let me walk you through a couple of clean, reusable approaches to handle monthly data counting and visualization with Pandas, no more manual sum calls needed!

First, Make Sure Your Dates Are Properly Typed #

Before anything else, you need to convert your string date column to a datetime type—this is critical for Pandas to recognize and group dates correctly:

import pandas as pd

# Convert your date column from string to datetime
part_date['date'] = pd.to_datetime(part_date['date'])

Method 1: Use resample() (Great for Time Series) #

Pandas' resample() is built exactly for aggregating time-based data. It’s super intuitive for monthly counts:

# Resample by month ('M' = end of month; use 'MS' for start of month if preferred)
monthly_counts = part_date.resample('M', on='date').size()

# If you set 'date' as your DataFrame index first, you can simplify it to:
# part_date = part_date.set_index('date')
# monthly_counts = part_date.resample('M').size()

The result will be a Series where the index is the end-of-month date, and values are the number of records for that month.

Method 2: Use groupby() with dt.to_period() #

If you prefer more explicit grouping (or want cleaner month labels like 2010-01 instead of full dates), use dt.to_period('M') to group by month-periods:

# Group by year-month periods and count records
monthly_counts = part_date.groupby(part_date['date'].dt.to_period('M')).size()

This gives you an index formatted as YYYY-MM, which is often easier to read in plots.

Visualization (Bar or Scatter Plot) #

Now let’s turn those counts into a plot. Matplotlib works perfectly here, and you can easily switch between bar charts (better for comparing volumes) or scatter plots (as you mentioned):

Bar Chart (Recommended for Frequency)

import matplotlib.pyplot as plt

plt.figure(figsize=(12, 6))
monthly_counts.plot(kind='bar')
plt.title('Monthly Data Record Volume')
plt.xlabel('Month')
plt.ylabel('Number of Records')
plt.xticks(rotation=45)  # Rotate labels so they don't overlap
plt.tight_layout()  # Fix label clipping
plt.show()

Scatter Plot (As Your Original Approach)

plt.figure(figsize=(12, 6))
plt.scatter(monthly_counts.index.astype(str), monthly_counts.values)
plt.title('Monthly Data Record Volume (Scatter Plot)')
plt.xlabel('Month')
plt.ylabel('Number of Records')
plt.xticks(rotation=45)
plt.tight_layout()
plt.show()

Reusable Function for Future Use #

To make this even easier, wrap everything into a reusable function—you can call it anytime with different DataFrames or date columns:

def generate_monthly_data_report(df, date_column='date', plot_type='bar'):
    # Convert date column to datetime
    df[date_column] = pd.to_datetime(df[date_column])
    # Calculate monthly counts
    monthly_counts = df.groupby(df[date_column].dt.to_period('M')).size()
    # Create plot
    plt.figure(figsize=(12, 6))
    if plot_type == 'bar':
        monthly_counts.plot(kind='bar')
    elif plot_type == 'scatter':
        plt.scatter(monthly_counts.index.astype(str), monthly_counts.values)
    # Format plot
    plt.title(f'Monthly Data Volume ({plot_type.capitalize()} Plot)')
    plt.xlabel('Month')
    plt.ylabel('Number of Records')
    plt.xticks(rotation=45)
    plt.tight_layout()
    plt.show()
    # Return the stats if you need to use them elsewhere
    return monthly_counts

# Example usage:
monthly_stats = generate_monthly_data_report(part_date, date_column='date', plot_type='scatter')

This way, you never have to write that repetitive hardcoded sum logic again—just pass in your data and go!

内容的提问来源于stack exchange，提问作者Rio