| Title |
Big Data Merging and Deep Learning Model Optimization for Improving Weather Information Forecasting Performance |
| Authors |
고경규(Kyung-Kyu Ko) ; 샤자드(Shahzad) ; 정은성(Eun-Sung Jung) |
| DOI |
https://doi.org/10.5573/ieie.2021.58.5.39 |
| Keywords |
Weather forecast; Time-series data; Deep learning; LSTM |
| Abstract |
Due to the rapid industrialization, many forms of air pollution are increasing, especially particulate matter has an adverse effect on the human body, such as causing or worsening heart and lung-related diseases. This research predicts particulate matter levels in 8-hours so that it can respond in advance as a solution to reduce fine dust damage. We construct various datasets for accurate prediction and utilize Stacked LSTM, LSTM AutoEncoder, and LSTM Variational AutoEncoder, models of the LSTM family that show good performance in time series data prediction. Research show that the increase in the number of data did not improve the performance of the model, but the increase in the number of features improved the performance of the model. Furthermore, the higher the number of features, the better the LSTM AutoEncoder performance. |