Time Series Visualization using Transformer for Prediction of Natural Catastrophe

Shivam Pandey, Mahek Jain

Abstract: The extension of the forecast time is an essential requirement for real-world applications, which includes early caution for severe climate conditions. In this paper, we come up with a new approach to time series forecasting. The time-series data is generic in lots of disciplines and engineering. Time series prediction is a vital assignment in time-series data modeling and is an important area of deep learning. We have developed a novel technique that makes use of a Transformer-based deep learning model for the prediction of time-series data. This technique works with the aid of self-attention mechanisms to study complicated patterns and dynamics from time-series data. Moreover, it is a preferred framework and may be implemented in univariate and multivariate time series data, in addition to time series embedding. Using natural disasters such as flood forecasting as a case study, we show that the forecast outcomes produced using our technique are similar to the state-of-the-art.

Keywords: Weather Forecasting, Transformer Networks, Time Series, Deep Learning, Attention Mechanisms