Model Evaluation on Air Pollutant Index (API) in Petaling Jaya, Malaysia
The measurement of air quality in Malaysia is described by Air Pollutant Index (API). The API value is computed based on average concentration of air pollutants specifically ozone, carbon monoxide, nitrogen dioxide, sulphur dioxide and particulate matter (diameter less than 10 microns). The air pollutant with the highest concentration will determine as the API value. This study aims to evaluate forecast accuracy by using monthly basis of Air Pollutant Index (API) dataset in Petaling Jaya, Malaysia. The forecasting methods used were included Naive, Naive with trend, Single Exponential Smoothing Technique, Double Exponential Smoothing Technique and Holt’s method. Furthermore, solver result was used to estimate optimal value of smoothing constant precisely to minimize the error measures for forecasting purpose. The error measures used to evaluate the model’s performance are Mean Square Error (MSE), Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). As a result, single exponential smoothing technique produced the smallest error measures compared to others. Hence, chosen as the most accurate forecast technique for one month ahead air pollution index (API) forecast. Conversely, single exponential smoothing model is most appropriate for non-seasonal trends with the absence of seasonal and cyclical variations. Further studies on Air Pollutant Index (API) or other five major air pollutants suggested to be conducted using an advanced forecasting methods such Box – Jenkins. Box – Jenkins approaches provide the most accurate short term forecast along with the use of most frequently data such as hourly or daily data.
Keywords: air pollutant index, forecasting, time series, Naive, Naive with trend, exponential smoothing technique, Holt’s method.