将字典中的精确匹配键映射到Pandas DataFrame列并返回对应值

Hey there! Let's tackle those keyword-to-value matching snags you're hitting with your Pandas DataFrame, building on the key-matching solution you already have. I’ll walk through the most common scenarios and fixes with concrete code examples:

1. Exact Value Matching #

If you need to match exact values between your keyword dictionary and DataFrame entries, the easiest approach is to reverse your dictionary to map values directly to their parent keys, then use Pandas' map() method.

Example Code: #

import pandas as pd

# Your existing keyword dictionary
keyword_dict = {"fruit": ["apple", "banana"], "veggie": ["carrot", "spinach"]}

# Sample DataFrame (adjust to match your actual data structure)
df = pd.DataFrame(
    {"items": ["apple", "orange", "carrot", "grape"], 
     "matched_category": ["", "", "", ""]}
)

# Reverse the dictionary to create a value-to-key lookup map
value_to_key = {v: k for key, values in keyword_dict.items() for v in values}

# Perform exact match and fill unmatched entries with "unknown"
df["matched_category"] = df["items"].map(value_to_key).fillna("unknown")

print(df)

Output: #

items matched_category
0    apple             fruit
1   orange           unknown
2   carrot             veggie
3    grape           unknown

2. Partial/Substring Matching #

If your DataFrame entries contain the keywords as substrings (e.g., "fresh apple" instead of just "apple"), use a custom function with apply() to check for substring matches. You can add case-insensitive matching to avoid missing matches due to capitalization.

Example Code: #

# Custom function to check for substring matches
def find_substring_match(item, keyword_map):
    # Convert item to lowercase for case-insensitive check
    lower_item = item.lower()
    for category, keywords in keyword_map.items():
        for keyword in keywords:
            if keyword.lower() in lower_item:
                return category
    return "unknown"

# Apply the function to your DataFrame column
df["matched_category"] = df["items"].apply(find_substring_match, keyword_map=keyword_dict)

# Test with a modified DataFrame containing substrings
df_test = pd.DataFrame({"items": ["Fresh Apple", "carrot soup", "orange juice"]})
df_test["matched_category"] = df_test["items"].apply(find_substring_match, keyword_map=keyword_dict)
print(df_test)

Output: #

items matched_category
0    Fresh Apple             fruit
1    carrot soup             veggie
2  orange juice           unknown

3. Handling Multiple Matches #

If a single DataFrame entry matches multiple keywords (e.g., "apple carrot salad"), modify the custom function to collect all matching categories instead of returning just one.

Example Code: #

def find_multiple_matches(item, keyword_map):
    lower_item = item.lower()
    matches = []
    for category, keywords in keyword_map.items():
        for keyword in keywords:
            if keyword.lower() in lower_item:
                matches.append(category)
    # Join matches with commas, or return "unknown" if none
    return ", ".join(set(matches)) if matches else "unknown"

# Test with a multi-match entry
df_multi = pd.DataFrame({"items": ["apple carrot salad", "spinach banana smoothie"]})
df_multi["matched_category"] = df_multi["items"].apply(find_multiple_matches, keyword_map=keyword_dict)
print(df_multi)

Output: #

items matched_category
0       apple carrot salad   fruit, veggie
1  spinach banana smoothie   fruit, veggie

Pro Tips for Debugging #

• Print your reversed value_to_key map first to confirm your keyword-to-category mappings are correct.
• Use df[df["matched_category"] == "unknown"] to isolate entries that aren't matching—check for typos, capitalization, or missing keywords in your dictionary.
• For large datasets, use regex-based matching with str.contains() for faster performance:

  # Build regex patterns for each category
  fruit_pattern = "|".join(keyword_dict["fruit"])
  veggie_pattern = "|".join(keyword_dict["veggie"])
  
  # Use np.where for vectorized matching
  import numpy as np
  df["matched_category"] = np.where(
      df["items"].str.contains(fruit_pattern, case=False), "fruit",
      np.where(df["items"].str.contains(veggie_pattern, case=False), "veggie", "unknown")
  )
  

内容的提问来源于stack exchange，提问作者J_Win