Weather Prediction Platform Using Data Mining

Abstract

This study focuses on using the Weka toolkit's machine learning algorithms to forecast weather conditions based on a dataset that includes columns for precipitation, temperature_max, temperature_min, and wind. Forecasting various weather conditions, such as drizzle, rain, sun, snow, and fog, is the goal. A thorough examination has been carried out utilizing a range of Weka algorithms, covering models with higher performance and lesser accuracy. After extensive testing, the best performing logistic algorithms meta. logitboost, meta. Iteration Classifier, trees.lmt, trees.j48, and functions. Logisticwere able to predict the weather with exceptional accuracy. These models demonstrated how well they could comprehend and predict weather conditions using the parameters from the given dataset. On the other hand, weather state predictions were relatively less accurate for models using the meta. classificationvia Regression, meta. bagging, trees.reptree, misc. input, and rules.zeror algorithms. In order to furnish stakeholders with practical and captivating insights, the outcomes were condensed and shown via user-friendly and aesthetically pleasing Google Looker dashboards. With the help of these dashboards, stakeholders may interactively examine and understand the predicted weather conditions that are based on the reliable machine learning algorithms. This research attempts to provide decision-makers with easily accessible, data-driven weather forecasts by combining these predictive algorithms with Google Looker's visualization capabilities. The dashboard system's extensive analysis and visualization are intended to improve the comprehension and use of machine learning in weather prediction, promoting well-informed decision-making across a range of industries dependent on weather-sensitive operations.