Let's break down your problem step by step, fix the hidden bugs in your original functions, and get your active_target working exactly how you want it.

First, Fix the active() Function #

Your original active function has two critical bugs that would cause errors before we even get to active_target:

1. You used groupby('customer_id') but your DataFrame’s column is named cust_id
2. After grouping to get the max order date, you lose the Frequency column, so the final filter on Frequency >= freq won’t work at all.

Here’s the corrected, functional version:

import pandas as pd
import datetime

# Your original DataFrame
df = pd.DataFrame({
    'cust_id': [1,1,2,2,1,3,3,4,4,3,3,3],
    'order_date': [
        '2015-01-16', '2019-12-03', '2014-12-21', '2015-01-10',
        '2015-01-10', '2018-01-18', '2017-03-04', '2019-11-05',
        '2010-01-01', '2019-12-06', '2002-01-01', '2018-01-01'
    ]
})
df['order_date'] = pd.to_datetime(df['order_date'])

def active(month_before_current_month, freq):
    # Calculate frequency per customer (use transform to keep all rows)
    df['Frequency'] = df.groupby('cust_id')['order_date'].transform('nunique')
    # Get most recent order date per customer
    recent_customers = df.groupby('cust_id')['order_date'].max().reset_index().rename(columns={'order_date': 'Most_recent_date'})
    # Merge back the frequency data we calculated earlier
    recent_customers = recent_customers.merge(df[['cust_id', 'Frequency']].drop_duplicates(), on='cust_id')
    # Calculate the cutoff date for "recent" transactions
    cutoff_date = datetime.date.today() - datetime.timedelta(days=month_before_current_month*365//12)
    # Filter customers who meet both criteria
    result = recent_customers[
        (recent_customers['Most_recent_date'].dt.date >= cutoff_date) & 
        (recent_customers['Frequency'] >= freq)
    ]
    return result

Calling active(1, 3) will now return your expected output correctly:

cust_id Most_recent_date  Frequency
0        1       2019-12-03          3
1        3       2019-12-06          5

Why Your Modified active_target() Throws an Error #

Your revised active_target uses **kwargs, which tells Python the function only accepts keyword arguments (like key=value pairs). But when you call active_target(active(1,3)), you’re passing a positional argument (the DataFrame returned by active()), which doesn’t match the function’s signature. That’s exactly what the error TypeError: active_target() takes 0 positional arguments but 1 was given means.

Two Fixes for active_target() #

Fix 1: Directly Accept the active() Result (Simplest) #

If your main goal is to pass the output of active() directly to active_target, just define the function to accept a DataFrame parameter explicitly:

def active_target(active_df):
    return active_df['Frequency'].mean()

# Call it like this:
active_target(active(1, 3))  # Returns (3+5)/2 = 4.0

Fix 2: Support Both Passing Parameters and Precomputed Data #

If you want flexibility (either pass parameters to let active_target run active() internally, or pass a precomputed DataFrame), adjust the function to handle both cases:

def active_target(**kwargs):
    # Check if we're given a precomputed active DataFrame
    if 'active_df' in kwargs:
        df = kwargs['active_df']
    else:
        # Extract parameters to run active() internally
        try:
            month = kwargs['month_before_current_month']
            freq = kwargs['freq']
            df = active(month, freq)
        except KeyError:
            raise ValueError("Either pass 'active_df' or both 'month_before_current_month' and 'freq'")
    return df['Frequency'].mean()

# Both calls work:
# 1. Pass precomputed data
active_target(active_df=active(1, 3))
# 2. Pass parameters to compute internally
active_target(month_before_current_month=1, freq=3)

Quick Notes #

• Always double-check column names when grouping/merging—small typos like customer_id vs cust_id are easy to miss but break everything.
• **kwargs is for keyword arguments only, not positional ones. If you want to accept arbitrary positional arguments, use *args, but in this case, an explicit parameter for the DataFrame is cleaner and more readable.

内容的提问来源于stack exchange，提问作者noob