FORECASTING POTATO PRODUCTION IN PAKISTAN USING BAYESIAN NON PARAMETRIC MODELLING Authors: Khurram H* and Iqbal MM
ABSTRACT
Forecasting and prediction provide basis for informed decision-making and their importance goes manifold particularly when they have macro impact covering a whole country. Quality of forecast directly affects the governance of a country. Statistics, if applied cautiously, have all the potential to do the job effectively. Forecasting and prediction in relation to the crop production is not new but getting Bayesian approach involved is relatively new and is rare for Pakistani environment let alone Bayesian Non-Parametric, which is the subject matter of the paper, is absent in this part of the world. Life long time-series data of Pakistan as a whole and on provincial basis was used for demonstration. Gaussian Process with an innovative kernel function was devised for the purpose. The performance of the proposed model was compared against existing well-taken models and the supremacy of the proposed model was established.
Keywords: Bayesian Nonparametric Modelling, Gaussian Process, Potato Production, Forecasting
