Double Exponential Smoothing in Forecasting the Numbers of Pole-Mounted Transformer Failures

In this paper, author studied the use of double exponential smoothing (DES) method to forecast the quarterly numbers of pole-mounted transformer failures. To determine the values of smoothing constants, they formulate a nonlinear mathematical programming (NLP) optimisation problem for this purpose, propose the use of an Excel Solver approach and exhaustive search-based method, with a program written in Python, are used to solve the optimisation. The two methods are used to find the optimal parameters of the forecast and results are compared. The exhaustive search-based approach delivers superior forecast results, with the forecasted values responding better with variation in observations and a lower forecast error obtained.