Handle data
manip is a class to unify various diagnostics methods and provide a consistent interface for diagnostics.
Perform one-hot encoding on specified columns
Args:
- cols (str or list): Columns to encode. Use ‘all’ for all object-type columns
Returns:
- pd.DataFrame: DataFrame with encoded columns
bi.dist.OHE(
self,
cols='all',
)Load data from CSV file
Args:
- path (str): Path to the CSV file
- **kwargs*: Additional arguments for pd.read_csv
Returns:
pd.DataFrame: Loaded dataframe
bi.dist.data(
self,
path,
**kwargs,
)Prepare data for model input in JAX format
Args:
- cols (list): List of columns to include in model data
Returns:
- dict: JAX formatted dictionary
bi.dist.data_to_model(
self,
cols,
)Create index encoding for categorical columns
Args:
- cols (str or list): Columns to encode. Use ‘all’ for all object-type columns
Returns:
- pd.DataFrame: DataFrame with encoded columns
bi.dist.index(
self,
cols='all',
)Convert pandas dataframe to JAX compatible format for a model
Args:
- model: JAX model to prepare data for
- bit (str): Bit precision for numbers (default: 32)
Returns:
- dict: JAX formatted dictionary
bi.dist.pd_to_jax(
self,
model,
bit=None,
)Standardize specified columns
Args:
- data (str or list): Columns to standardize. Use ‘all’ for all columns
Returns:
- pd.DataFrame: Standardized dataframe
bi.dist.scale(
self,
data='all',
)JAX-jitted function to scale/standardize a single variable
bi.dist.scale_var(
self,
x,
)Convert specified columns to float type
Args:
- cols (str or list): Columns to convert. Use ‘all’ for all columns
- type (str): Float type to convert to (default: float32)
Returns:
- pd.DataFrame: Converted dataframe
bi.dist.to_float(
self,
cols='all',
type='float32',
)Convert specified columns to integer type
Args:
- cols (str or list): Columns to convert. Use ‘all’ for all columns
- type (str): Integer type to convert to (default: int32)
Returns:
- pd.DataFrame: Converted dataframe
bi.dist.to_int(
self,
cols='all',
type='int32',
)