Scholarlink

There is a wide range of applications for weather forecasting systems (WFS), and it is high time to improve their performance. This paper initially validates the basic WFS code and then proposes the use of an artificial neural network (ANN)-based data prediction system. The proposal suggests creating a reduced dataset by eliminating not a number values (NANs) from the datasets. The ultimate goal of the paper is to maximize the R-squared value to better represent the correlation between predicted and actual data. Performance is also evaluated based on the Mean Squared Error (MSE) estimation, with the proposed system using 19 inputs and 14 hidden layers. The best possible training stage achieves an R-squared value of 0.904, and the minimum MSE is recorded as 0.3962 for Delhi and 0.2624 for Mumbai data, indicating successful training phase performance.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Exploring the Effectiveness of Artificial Neural Networks and Regression Models in Weather Prediction

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Vishwadeep Singh,
  • Chandan Kumar,
  • Nitin Choudhary

摘要

There is a wide range of applications for weather forecasting systems (WFS), and it is high time to improve their performance. This paper initially validates the basic WFS code and then proposes the use of an artificial neural network (ANN)-based data prediction system. The proposal suggests creating a reduced dataset by eliminating not a number values (NANs) from the datasets. The ultimate goal of the paper is to maximize the R-squared value to better represent the correlation between predicted and actual data. Performance is also evaluated based on the Mean Squared Error (MSE) estimation, with the proposed system using 19 inputs and 14 hidden layers. The best possible training stage achieves an R-squared value of 0.904, and the minimum MSE is recorded as 0.3962 for Delhi and 0.2624 for Mumbai data, indicating successful training phase performance.