A Model for Weather Prediction Using Hybrid Machine Learning Algorithms

Abstract

Weather prediction is crucial in agriculture, transportation, and disaster management. Accurate and timely forecasts are essential for making informed decisions and minimizing risks. This study proposes a hybrid machine learning approach for weather prediction, combining Ridge Regression for feature extraction and selection and Long Short-Term Memory (LSTM) for training and evaluation. Ridge Regression extracts and selects relevant features from input data using Python and libraries like scikit-learn, Pandas, and NumPy. The selected features are then fed into the LSTM model, a recurrent neural network (RNN) capable of capturing temporal dependencies and patterns in time series data. The hybrid model achieves superior accuracy in weather prediction compared to individual algorithms, with experimental results of 3.38% in mean absolute error, 0.2% in RMSE, and 4.7% in MAPE. The findings highlight the potential of hybrid machine learning algorithms in improving weather prediction models, contributing to better decision-making in sectors dependent on weather information. Further research can explore additional hybrid approaches and evaluate their performance on larger and more diverse datasets, paving the way for advancements in weather prediction. prediction systems.