imputegap.algorithms.mrnn package

The imputegap.algorithms.mrnn package contains various imputation algorithms used for handling missing values in time series data.

Submodules

Modules

imputegap.algorithms.mrnn.mrnn(incomp_data, hidden_dim, learning_rate, iterations, sequence_length, logs=True)[source]

Perform imputation using the Multivariate Recurrent Neural Network (MRNN) algorithm.

Parameters

incomp_datanumpy.ndarray

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

hidden_dimint

The number of hidden dimensions in the MRNN model.

learning_ratefloat

The learning rate for the training process.

iterationsint

The number of iterations for training the MRNN model.

sequence_lengthint

The length of sequences used within the MRNN model.

logsbool, optional

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

Returns

numpy.ndarray

The imputed matrix with missing values recovered.

Notes

The MRNN algorithm is a machine learning-based approach for time series imputation, where missing values are recovered using a recurrent neural network structure.

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

Example

>>> recov_data = mrnn(incomp_data, hidden_dim=64, learning_rate=0.001, iterations=1000, sequence_length=7)
>>> print(recov_data)

References

  1. Yoon, W. R. Zame and M. van der Schaar, “Estimating Missing Data in Temporal Data Streams Using Multi-Directional Recurrent Neural Networks,” in IEEE Transactions on Biomedical Engineering, vol. 66, no. 5, pp. 1477-1490, May 2019, doi: 10.1109/TBME.2018.2874712. keywords: {Time measurement;Interpolation;Estimation;Medical diagnostic imaging;Correlation;Recurrent neural networks;Biomedical measurement;Missing data;temporal data streams;imputation;recurrent neural nets}