The daily number of calls received by a call center over the last three weeks is reported by the Assignment 2.xlsx spreadsheet on Blackboard. The daily data is further broken down into four time windows. Please,
(i) Graph the 3×7×4 = 84 raw data points, and describe what you see using the concepts of trend, seasonality and randomness/noise.
(ii) Find the four seasonal indexes for the time of the day, and the seven seasonal indexes for the days of the week. Briefly describe what they represent and how they can be used in a forecast.
(iii) Next to the raw data point column, create a second column where each data point is divided by its corresponding "time of the day" seasonal index, and create a third column where each data point is divided by its corresponding "time of the day" indexand"day of the week" index. Graph the four columns and describe what you see. What does dividing a data point by a seasonal index do?
(iv) On the raw data point column, use a non-weighted 10-period moving average to forecast the number of calls for the next two days. Try to use the seasonal indexes found in (ii) to account for the "time window" and "day of the week" seasonal factors.
(v) Suppose that over the last two weeks you have been testing a non-weighted 10-period moving average and a non-weighted 4-period moving average. For the future, you want to choose whichever model has performed better over such testing period. Please, make your choice and motivate it using the forecasting errors as a guidance.