pandas-schema

0.3.6 verified Fri May 01 auth: no python maintenance

A validation library for Pandas data frames using user-friendly schemas. Current version is 0.3.6, with infrequent releases.

pip install pandas-schema

Common errors

error ImportError: cannot import name 'Check' from 'pandas_schema' ↓

cause Check class is in the validation submodule, not at package level.

fix

Use: from pandas_schema.validation import Check

error AttributeError: 'DataFrame' object has no attribute 'validate' ↓

cause Confusion with pandas built-in validate; the package provides schema.validate(df).

fix

Instantiate DataFrameSchema and call schema.validate(df).

Warnings

gotcha Column type parameter expects Python types (e.g., int, float) not numpy dtypes. Using 'int64' may cause unexpected behavior. ↓

fix Use int, float, str, etc.

gotcha Validators are in pandas_schema.validation; importing from pandas_schema directly won't give Check or validation classes. ↓

fix Use from pandas_schema.validation import Check, CanConvertValidation, etc.

gotcha InRangeValidation crashes on non-numeric text in versions <=0.3.5. This is fixed in 0.3.6. ↓

fix Upgrade to 0.3.6 or ensure data is numeric before validation.

Imports

DataFrameSchema

wrong

from pandas_schema.schema import DataFrameSchema

correct

from pandas_schema import DataFrameSchema

DataFrameSchema is exposed at the package level

Column
```
from pandas_schema import Column
```

Check

wrong

from pandas_schema import Check

correct

from pandas_schema.validation import Check

Check and validators are in the validation submodule

Quickstart

Define a schema and validate a DataFrame. Returns list of errors.

import pandas as pd
from pandas_schema import DataFrameSchema, Column
from pandas_schema.validation import CanConvertValidation

schema = DataFrameSchema({
    'age': Column(int, [
        CanConvertValidation(int)
    ])
})
df = pd.DataFrame({'age': ['25', '30']})
errors = schema.validate(df)
if errors:
    for error in errors:
        print(error)
else:
    print('Validation passed')