facultyai / marshmallow-dataframe Goto Github PK

Hello, thank you very much for your project, I've been looking for a more practical way to do a schema treatment with pandas.

I would like some help, in my project, separated schema into a module and instantiate, for example, the class DataSchema (SplitDataFrameSchema). But she dont recognize Meta parameter, dtype, how can I dynamically assign it (based on the input data)?

My structure in Tree:

├── app
│   ├── api.py
│   ├── __init__.py
│   ├── db
│   │   ├── core.py
│   │   ├── __init__.py
│   ├── resources
│   │   ├── analysis_data.py
│   │   ├── __init__.py
│   ├── schemas.py
│   └── utils.py
├── config.py
├── connections.log
└── main.py

My Schema code:

from marshmallow_dataframe import SplitDataFrameSchema

class DataSchema(SplitDataFrameSchema):
    """Automatically generated schema for dataframe"""

    class Meta:
        dtypes = self.dtypes

Hi, and thanks for the work on this package !
I am also trying to get that working nicely together...

I recently encountered :

{'index': ['Missing data for required field.']}

But I expected index to be optional (when not present assume range(n), same as pandas.DataFrame()

What do you think ?

Right now categorical columns are always recognised as a string dtype, which could be wrong. In addition, we are not validating that the deserialised values are one of the category options.