load_from_tsfile¶
- load_from_tsfile(full_file_path_and_name, replace_missing_vals_with='NaN', return_meta_data=False, return_type='auto')[source]¶
Load time series .ts file into X and (optionally) y.
- Parameters:
- full_file_path_and_namestring
full path of the file to load, .ts extension is assumed.
- replace_missing_vals_withstring, default=”NaN”
issing values in the file are replaces with this value
- return_meta_databoolean, default=False
return a dictionary with the meta data loaded from the file
- return_typestring, default = “auto”
data type to convert to. If “auto”, returns numpy3D for equal length and list of numpy2D for unequal. If “numpy2D”, will squash a univariate equal length into a numpy2D (n_cases, n_timepoints). Other options are available but not supported medium term.
- Returns:
- data: Union[np.ndarray,list]
time series data, np.ndarray (n_cases, n_channels, n_timepoints) if equal length time series, list of [n_cases] np.ndarray (n_channels, n_timepoints) if unequal length series.
- ytarget variable, np.ndarray of string or int
- meta_datadict (optional).
dictionary of characteristics, with keys “problemname” (string), booleans: “timestamps”, “missing”, “univariate”, “equallength”, “classlabel”, “targetlabel” and “class_values”: [],
- Raises:
- IOError if the load fails.