imputegap.algorithms.iim package¶

Submodules¶

Modules¶

imputegap.algorithms.iim.iim(incomp_data, number_neighbor, algo_code, logs=True)[source]¶

Perform imputation using the Iterative Imputation Method (IIM) algorithm.

Parameters¶

incomp_datanumpy.ndarray

The input matrix with contamination (missing values represented as NaNs).

number_neighborint

The number of neighbors to use for the K-Nearest Neighbors (KNN) classifier (default is 10).

algo_codestr

The specific action code for the IIM output. This determines the behavior of the algorithm.

logsbool, optional

Whether to log the execution time (default is True).

Returns¶

numpy.ndarray

The imputed matrix with missing values recovered.

Notes¶

The IIM algorithm works by utilizing K-Nearest Neighbors (KNN) to estimate missing values in time series data. Depending on the provided algo_code, different versions of the algorithm may be executed.

The function logs the total execution time if logs is set to True.

Example¶

>>> recov_data = iim(incomp_data, number_neighbor=10, algo_code="iim 2")
>>> print(recov_data)

References¶

A. Zhang, S. Song, Y. Sun and J. Wang, “Learning Individual Models for Imputation,” 2019 IEEE 35th International Conference on Data Engineering (ICDE), Macao, China, 2019, pp. 160-171, doi: 10.1109/ICDE.2019.00023. keywords: {Data models;Adaptation models;Computational modeling;Predictive models;Numerical models;Aggregates;Regression tree analysis;Missing values;Data imputation}