Quantile smoothing splines regression on the air relative humidity toward the rubber production in South Sumatera

Rubber is one of the Indonesian plantation commodities which has an important role in economic activity. Rubber can grow very well in the tropical area. One of the factors influencing the rubber production is the air relative humidity. This research is aimed finding out the influence of the air relative humidity and the rubber production in South Sumatera. The relationship pattern shows different shapes in some parts indicate a difficulty in specified parametrically, then a nonparametric approach was proposed. In this case we use smoothing splines technique. However, the pattern spreads asymmetrically and contains outliers. Hence, we also propose to use a quantile objective function called quantile smoothing splines. In this research, we divide the data distribution into five parts based on very low, low, moderate, high, and very high characteristics. The performances of quantile smoothing splines (with τ = 0.5, as median case) and smoothing splines (mean approach) are compared using Root Mean Square Error. The result has shown that the quantile smoothing splines regression gives smaller Root Mean Square Error compared to smoothing splines regression (mean approach).