Weather Prediction Model using Ridge Regression

Weather prediction is crucial for various sectors such as agriculture, transportation, and disaster management. Accurate forecasts enable proactive measures to mitigate risks and optimize operations. Traditional methods rely on complex numerical models, while machine learning (ML) offers a promising alternative due to its ability to provide accurate predictions with less computational overhead. This study focuses on predicting the next day’s maximum temperature using ridge regression models applied to historical weather data from Bangalore, India. The research emphasizes feature engineering techniques to enhance model performance, including the creation of rolling window statistics and expanding averages of temperature data. The linear regression model is replaced with ridge regression due to its capability to handle multicollinearity and overfitting issues. Evaluation metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared ($\mathbf{R}^{2}$) are employed to assess model accuracy. The results demonstrate that the ridge regression model achieves promising accuracy in predicting daily maximum temperatures. This paper discusses the methodology, experiments, and results in detail, highlighting the model’s performance.