aixd.data.utils_data.convert_to

aixd.data.utils_data.convert_to(data: pd.DataFrame | np.ndarray | List[List] | Dict | List[Dict] | torch.Tensor, format: str = 'df', dataobjects: List[DataObject] = None) → pd.DataFrame | np.ndarray | List[List] | Dict | List[Dict] | torch.Tensor[source]

Takes any data format, detect the type, and convert it if neccessary.

The possible formats are:

  • “dict” : dictionary

  • “dict_list” : list of dictionaries

  • “df_per_obj” : dataframe

  • “df” : standard dataframe flattened, all cells contain single values and column names are renamed.

  • “array” : nparray

  • “torch” : torch tensor

  • “list” : nested list

In principle, the only inputs are the data, the desired format and the data objects. This is all required to detect and then convert

Important: this functions only works with the data. This means it does not expect any “uid” or “error” columns. These will cause errors.

Parameters:

  • data (Union[pd.DataFrame, np.ndarray, List[List], Dict, List[Dict], torch.Tensor]) – Data with some specific format.

  • format (str, optional, default=”df”) – Desired target format.

  • dataobjects (List[DataObject], optional, default=None) – List of data objects from which names and dimensions will be recovered.

Returns:

data_out (Union[pd.DataFrame, np.ndarray, List[List], Dict, List[Dict], torch.Tensor]) – Converted data

On this page