Weather forecasting using Convex hull & K-Means Techniques An Approach

TL;DR

This paper introduces a novel weather forecasting method that combines convex hulls for structural data representation with incremental K-Means clustering to predict weather conditions in Kolkata based on pollution and time series data.

Contribution

The paper presents a new approach integrating convex hulls and incremental K-Means clustering for weather prediction using non-structural time series data.

Findings

Effective clustering of weather data into four categories

Improved prediction accuracy using the combined technique

Application to Kolkata's weather data demonstrates feasibility

Abstract

Data mining is a popular concept of mined necessary data from a large set of data. Data mining using clustering is a powerful way to analyze data and gives prediction. In this paper non structural time series data is used to forecast daily average temperature, humidity and overall weather conditions of Kolkata city. The air pollution data have been taken from West Bengal Pollution Control Board to build the original dataset on which the prediction approach of this paper is studied and applied. This paper describes a new technique to predict the weather conditions using convex hull which gives structural data and then apply incremental K-means to define the appropriate clusters. It splits the total database into four separate databases with respect to different weather conditions. In the final step, the result will be calculated on the basis of priority based protocol which is defined based on some mathematical deduction.