Automatic Data Imputation in Time Series Processing Using Neural Networks for Industry and Medical Datasets
8th Annual International Conference on Information Management and Big Data, SIMBig 2021
; 1577 CCIS:3-16, 2022.
Article
in English
| Scopus | ID: covidwho-1826267
ABSTRACT
Time series classification and regression techniques help solve problems in many knowledge areas, including medicine, electronics, industry, and even music. When we apply them to real-life issues, a common obstacle is the lack of data in intervals within a time series. Usually, to solve it, the missing data is populated with information highly dependent on available datasets, which requires prior analysis. This paper addresses the problem in a novel way, automatically filling the missing data using a mixture of techniques and letting the prediction model decide which filling is better. We tested our approach for classification in industrial and medical datasets and for regression, we used a dataset containing COVID-19 information. Our results are very competitive, and our approach improves the state-of-the-art models. We obtain better performance in all the experiments for the selected quality measures. Most importantly, the improvement is more statistically significant when the amount of missing data is higher. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
Convolutional networks; Deep learning; Long short-term memory; Regression; Time series classification; Convolution; Convolutional neural networks; Electronics industry; Regression analysis; Time series; Classification technique; Data imputation; Medical data sets; Missing data; Neural-networks; Time series classifications; Time series processing; Time-series regression; Classification (of information)
Full text:
Available
Collection:
Databases of international organizations
Database:
Scopus
Type of study:
Experimental Studies
Language:
English
Journal:
8th Annual International Conference on Information Management and Big Data, SIMBig 2021
Year:
2022
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS