Determination of Chennai Annual Rainfall Using Datamining Technique
DOI:
https://doi.org/10.9734/bpi/ramrcs/v8/1718BKeywords:
Rainfall prediction, data mining, k-nearest neighbor, inverse distance weighting and MAPEAbstract
India is basically an agricultural country and weather forecasting is a challenge that has depended primarily on model-based methods. A basic requirement for human survival is water and rainfall is one of the main source of water. Rainfall prediction is very vital for agricultural and meteorological department and is one of the most challenging tasks. Due to the increase in population, the demand for water has also increased. This has resulted in decrease of availability of water as the ground water levels have been depleted. Chennai city in particular depends heavily on its ground water resources which are replenished by rainwater. Thus, early prediction of rainfall is very much vital. Various techniques like ARIMA, ANN, Regression analysis, Genetic Algorithm, Fuzzy logic, SVM etc., are applied for rainfall prediction. This paper presents one of the data mining technique k-nearest neighbour for the prediction of Chennai rainfall. Validation of the model is done by the error measure Mean Absolute Percentage Error (MAPE) and analysis show that k-nearest neighbour for k = 3 yielded best results when compared to k = 5.
