基于Pandas实现交易数据集购买商品X的客户标签分类

First, let's align on the exact tagging rules we're implementing:

• great: Customer had purchases before their first purchase of product X
• boo: Customer only purchased product X exactly once (no other purchases)
• awesome: Customer's first purchase was X, and they bought other products afterward

I'll use pandas for this since it's the standard tool for working with tabular transaction data. Here's a practical, tested implementation:

Step 1: Sample Test Data #

Let's start with a dataset that matches your example customer scenarios:

import pandas as pd

# Sample transaction records
data = {
    'custID': [1, 1, 1, 2, 3, 4, 4, 4],
    'product': ['A', 'X', 'B', 'X', 'X', 'X', 'C', 'D'],
    'transaction_date': ['2023-01-01', '2023-01-05', '2023-01-10', 
                         '2023-02-01', '2023-03-01', '2023-04-01', 
                         '2023-04-05', '2023-04-10']
}
df = pd.DataFrame(data)

Step 2: Tagging Function #

This function groups transactions by customer, sorts them by date to preserve purchase order, then applies your tagging logic:

def tag_customers(df, target_product='X'):
    # Sort transactions to ensure we process purchases in chronological order
    sorted_transactions = df.sort_values(['custID', 'transaction_date'])
    
    # Group data by customer ID to process each customer's history separately
    customer_groups = sorted_transactions.groupby('custID')
    
    tag_results = []
    
    for cust_id, group in customer_groups:
        # Get the sequence of products the customer bought (in order)
        purchase_sequence = group['product'].tolist()
        
        # Handle edge case: customer never bought the target product (just in case)
        if target_product not in purchase_sequence:
            tag_results.append({'custID': cust_id, 'tag': 'no_target_purchase'})
            continue
        
        first_purchase = purchase_sequence[0]
        
        if first_purchase != target_product:
            # Customer had existing purchases before buying X
            tag_results.append({'custID': cust_id, 'tag': 'great'})
        else:
            # First purchase was X—check if they bought other products later
            has_subsequent_non_target = any(p != target_product for p in purchase_sequence[1:])
            
            if has_subsequent_non_target:
                tag_results.append({'custID': cust_id, 'tag': 'awesome'})
            else:
                # All purchases are X—check if it's exactly one purchase
                if len(purchase_sequence) == 1:
                    tag_results.append({'custID': cust_id, 'tag': 'boo'})
                else:
                    # Edge case: multiple purchases of X, no other products
                    tag_results.append({'custID': cust_id, 'tag': 'multiple_x_only'})
    
    # Convert results to a DataFrame for easy integration with your original data
    return pd.DataFrame(tag_results)

Step 3: Test the Function #

Run the function on our sample data to verify it works as expected:

customer_tags = tag_customers(df)
print(customer_tags)

Output: #

custID              tag
0       1            great
1       2              boo
2       3              boo
3       4          awesome

This perfectly matches your example cases!

Edge Case Note #

I added handling for a scenario you didn't explicitly mention: customers who bought X multiple times but no other products (tagged as multiple_x_only). You can adjust this tag to whatever makes sense for your use case, or remove this branch if you don't expect this scenario in your dataset.

内容的提问来源于stack exchange，提问作者spr_m