Research on CNN (Convolutional Neural Network)-based Forecasting Method
DOI:
https://doi.org/10.30837/csitic52021231823Keywords:
time series, forecasting, convolutional neural networks, InceptionTime, COVID-19Abstract
The article deals with the problem of predicting the COVID-19 spreading in Ukraine. This task becomes more relevant each day. From all the modern models of CNN-based time series forecasting, the InceptionTime model was chosen. Its advantages are high accuracy and scalability. The research is implemented using the Python high-level programming language. The results are presented as the mean absolute error for each region. The InceptionTime model has proved its high accuracy for this task.
References
Y. Andrusenko «Analysis of Basic Time Series Forecasting Models» Science and Technology of the Air Force of Ukraine № 3(65), 2020, pp.91–96.
Hassan Ismail Fawaz, Benjamin Lucas, Germain Forestier. InceptionTime: Finding AlexNet for Time Series Classificatio. Data Mining and Knowledge Discovery. 2020. Vol. 34, p.1936–1962.
