如何在Pandas透视表中显示分类标签而非数值编码

Got it, this is a super common pain point when working with encoded categorical datasets—luckily, Pandas has straightforward ways to swap those numeric codes for human-readable labels before (or even while) building your pivot table. Let’s walk through this step by step:

First, translate your codebook into a Python dictionary (or nested dictionary if you have multiple encoded columns). For example:

# Nested dict for multiple columns
encoding_mapping = {
    "gender": {1: "Male", 2: "Female"},
    "marital_status": {0: "Single", 1: "Married", 2: "Divorced"},
    "education_level": {1: "High School", 2: "Bachelor's", 3: "Master's", 4: "PhD"}
}

# Or a single dict for one column if that's all you need
gender_mapping = {1: "Male", 2: "Female"}

You have two flexible options here depending on whether you want to modify your original DataFrame or not:

Option A: Modify the Original DataFrame (Permanent Change) #

Use df.replace() to swap codes for labels across all relevant columns in one go:

# Apply mapping to all columns in the encoding dict
df.replace(encoding_mapping, inplace=True)

Or target a single column with map():

# Update just the gender column
df["gender"] = df["gender"].map(gender_mapping)

Important note: map() will turn any unmapped numeric values into NaN. If you want to keep those original values instead, use fillna():

df["gender"] = df["gender"].map(gender_mapping).fillna(df["gender"])

Option B: Keep Original DataFrame Intact (Temporary Mapping) #

If you don’t want to alter your raw data, you can apply the mapping directly within the pivot_table() call:

pivot = df.pivot_table(
    values="monthly_salary",  # Your numeric value column
    index=df["gender"].map(gender_mapping),  # Map codes to labels for rows
    columns=df["education_level"].map(encoding_mapping["education_level"]),  # Map for columns
    aggfunc="mean"  # Your desired aggregation (sum, count, etc.)
)

Once your codes are replaced with labels, creating the pivot table works exactly like normal—except now the rows/columns will show human-readable labels instead of numbers. For example:

# If you modified the original DataFrame
final_pivot = df.pivot_table(
    values="monthly_salary",
    index="gender",
    columns="education_level",
    aggfunc=["mean", "count"],
    margins=True
)

This will give you a pivot table with "Male"/"Female" instead of 1/2, and education labels instead of numeric codes—exactly what you’re looking for!

内容的提问来源于stack exchange，提问作者B.Poe