Comparative analysis of Naïve Bayes , K-Nearest Neighbor and K-Means Clustering Method in Weather Forecast
Yulian Findawati1, Ika Ratna Indra Astutik2, Arif Senja Fitroni3, Indrawati4
Universitas Muhammadiyah Sidoarjo
Abstract
Weather forecast in an area is unpredictable. This is due to the fact that human factors cannot predict it. However, by presenting previous weather reports or data, science can be used as a reference to see weather forecast patterns . The weather forecast is by applying data mining using the algorithm Naive Bayes , K-nearest Neighbor (K-NN), and K-Means Clustering. Bayesian Classification is a statistical classification method that is useful for the process of determining the probability of a class membership. K-Nearest Neighbor Algorithm is a classification algorithm based on the similarity between one data and another data. K-Means clustering is an unsupervised learning algorithm that, as the name hints, finds a fixed number (k) of clusters in a set of data. Data used in the last 3 years (starting 2014-2017) which is daily weather data from the website of the Meteorology, Climatology and Geophysics Agency. Daily weather conditions consist of rainfall, irradiation, maximum temperature, minimum temperature, maximum humidity, minimum humidity, maximum air pressure, minimum air pressure, wind speed and wind direction. While the weather forecast label is not rainy, light rain, moderate rain, heavy rain and very heavy rain. Analysis using Rapid Miner tools
Keywords: Weather Forecast, Naïve Bayes, K-Nearest Neighbour, K-Means Clustering
Topic: Computer Science