paddlets.transform.fill

class Fill(cols: Union[str, List[str]], method: str = 'pre', value: int = 0, window_size: int = 10, min_num_non_missing_values: int = 1)[source]

Bases: BaseTransform

This class is designed to fill missing values in columns. There are three kinds of ways to fulfill this task, including

Replace the missing values with a statistic computed from a sliding window, e.g. MAX, MIN, MEAN, or MEDIAN;

Replace the missing values with adjacent values, which could be values previous or next to the missing values;

Replace the missing values with the value specified by the user.

Parameters

  • cols (str|List) – Column name(s) to be processed.

  • method (str) – Method of filling missing values. Totally 8 methods are supported currently: max: Use the max value in the sliding window. min: Use the min value in the sliding window. mean: Use the mean value in the sliding window. median: Use the median value in the sliding window. pre: Use the previous value. next: Use the next value. zero: Use 0s. default: Use the value specified by the user.

  • value (int||float) – Only effective when the method is default, value specified by the user to replace the missing values.

  • window_size (int) – Size of the sliding window.

  • min_num_non_missing_values (int) – Minimum number of non-missing values in the sliding window, if less than the min_num_non_missing_values, the statistic will be set to np.nan.

Returns

None

fit_one(dataset: TSDataset)[source]
Parameters

dataset (TSDataset) – dataset to process

Returns

self

transform_one(dataset: TSDataset, inplace: bool = False) → TSDataset[source]

Fill missing values.

Parameters

  • dataset (TSDataset) – TSDataset or List[TSDataset]

  • inplace (bool) – Set to True to perform inplace row normalization and avoid a copy.

Returns

Transformed TSDataset.

Return type

new_ts(TSDataset)