You are provided with four years of monthly demand data for a chemical component produced by a major Energy Company (see Quiz 5.XLS). By plotting the data visually examine the time series to identify any component such as trend, seasonality and structural change. Using the first three years, optimize SES based on MSE. Then, simulate forecasting for the last year of data by applying MA(3) and SES with alpha parameter obtained earlier. Which forecasting method do you suggest to be adopted, SES or MA(3)?

Question

You are provided with four years of monthly demand data for a chemical component produced by a major Energy Company (see Quiz 5.XLS). By plotting the data visually examine the time series to identify any component such as trend, seasonality and structural change. Using the first three years, optimize SES based on MSE. Then, simulate forecasting for the last year of data by applying MA(3) and SES with alpha parameter obtained earlier. Which forecasting method do you suggest to be adopted, SES or MA(3)?

Answer 1

Based on the information provided, I would suggest using the Seasonal Exponential Smoothing (SES) method for forecasting the last year of data. This is because the data has a strong seasonal pattern, and SES is specifically designed to handle time series data with seasonal variations.To optimize the SES method based on Mean Squared Error (MSE), you can follow these steps:1. Plot the time series data to visually examine the trend, seasonality, and structural change.2. Divide the data into training and testing sets. Use the first three years of data as the training set and the last year as the testing set.3. Apply the SES method to the training set and tune the alpha parameter to minimize the MSE between the forecasted and actual values.4. Once you have optimized the alpha parameter, apply the SES method to the testing set to forecast the last year of data.To simulate forecasting using the Moving Average (MA) method with a 3-period window, you can follow these steps:1. Calculate the moving average for each month in the last year of data. This can be done by taking the average of the current month and the previous two months.2. Apply the moving average method to forecast the next month's demand.In general, if the data has a strong seasonal pattern, SES is a more appropriate method than MA. However, if the seasonal pattern is weak or if there are structural changes in the data, MA may be more suitable. It is important to evaluate the performance of both methods and choose the one that provides the best forecast accuracy for your specific application.

Soru

Çözüm

Cevap