Grasia / knnp

Time Series Forecasting using K-Nearest Neighbors Algorithm (Parallel approach)

knnp : Time Series Prediction using K-Nearest Neighbors Algorithm (Parallel)

First release was developed as an End-of-Degree Project.

Further improvements have been made now as a project from GRASIA investigation group: https://grasia.fdi.ucm.es/

Purpose

This package intends to provide R users or anyone interested in the field of time series prediction the possibility of aplying the k-nearest neighbors algorithm to time series prediction problems. Two main functionalities are provided:

• Time series prediction using this method.
• Optimization of parameteres k and d of the algorithm.

All the code involved has been optimized to:

• Parallelize critic components as the process of optimization of parameteres k and d or the calculation of distances.
• Use memory efficiently.

Authors

• Daniel Bastarrica Lacalle
• Javier Berdecio Trigueros

Directors

• Javier Arroyo Gallardo
• Albert Meco Alias

Maintainer

• Daniel Bastarrica Lacalle

License

AGPL-3