Research Article
Time Series Prediction with Preprocessing and Clustering
@INPROCEEDINGS{10.1007/978-3-030-94182-6_27, author={Haoxuan Sun and Shuai Lin and Lin Han and Jidong Feng and Mingxu Sun}, title={Time Series Prediction with Preprocessing and Clustering}, proceedings={IoT and Big Data Technologies for Health Care. Second EAI International Conference, IoTCare 2021, Virtual Event, October 18-19, 2021, Proceedings, Part II}, proceedings_a={IOTCARE PART 2}, year={2022}, month={6}, keywords={Time series Kmeans clustering Similarity}, doi={10.1007/978-3-030-94182-6_27} }- Haoxuan Sun
Shuai Lin
Lin Han
Jidong Feng
Mingxu Sun
Year: 2022
Time Series Prediction with Preprocessing and Clustering
IOTCARE PART 2
Springer
DOI: 10.1007/978-3-030-94182-6_27
Abstract
This paper studies the similarity of time series, and studies the influence of weight on prediction results on the basis of clustering. We first introduce the practical significance and research purpose of the selected topic, summarizes the current research situation at home and abroad, and summarizes the research content of this paper. Second, we describe related concepts. Later, based on Dodger data set, we study the flow of total prediction data of time series. First of all, feature extraction of the data, pre-processing work, the original data generation time series. Then the data are processed and divided into training data and test data for the convenience of subsequent processing. Then the clustering algorithm was used to divide the time series into categories, and seven categories were divided according to the characteristics of one week time cycle. The average value of each category is calculated to replace the characteristics of the current category, and then the similarity is compared. Finally, the weight of each category is calculated by similarity degree, and then the data is predicted. MAE, R-squared, MAPE and other indicators were used to analyze and evaluate the forecast data.
