Rain Prediction Based on Machine Learning

DOI

10.2991/assehr.k.220504.536How to use a DOI?

Keywords

machine learning; Rain prediction; LSTM

Abstract

Our purpose is to try to use machine learning algorithms to predict the weather of the next day, since whether it will rain tomorrow is a very important indicator of the weather. In order to find the most predictable attributes of rain, The researcher use line charts, matrix graphs, and scatterplot graphs for visualization and analysis. The researcher find that several pairs of attributes have a high degree of similarity and correlation. In the fitting stage, the researcher used simple models such as KNN, decision tree, and ridge regression to evaluate its basic prediction quality and found that the accuracy rate is around 0.78. Since in the visualization stage, the researcher found that the samples that rained today have a slightly higher probability of raining the next day, the researcher tried to use LSTM to analyze the impact of historical weather and found that the relationship is not strong. Finally, logistic regression turns out to have the highest accuracy of 0.85, followed by adaboost with an accuracy of 0.82. Whether it will rain remains unpredictable to some extent.

Copyright

© 2022 The Authors. Published by Atlantis Press SARL.

Open Access

This is an open access article distributed under the CC BY-NC 4.0 license.

Download article (PDF)

TY  - CONF
AU  - Ye Zhao
AU  - Hanqi Shi
AU  - Yifei Ma
AU  - Mengyan He
AU  - Haotian Deng
AU  - Zhou Tong
PY  - 2022
DA  - 2022/06/01
TI  - Rain Prediction Based on Machine Learning
BT  - Proceedings of the 2022 8th International Conference on Humanities and Social Science Research (ICHSSR 2022)
PB  - Atlantis Press
SP  - 2957
EP  - 2970
SN  - 2352-5398
UR  - https://doi.org/10.2991/assehr.k.220504.536
DO  - 10.2991/assehr.k.220504.536
ID  - Zhao2022
ER  -