Date	Sales (000,000)
Sep-12	206
Oct-12	245
Nov-12	185
Dec-12	169
Jan-13	162
Feb-13	177
Mar-13	207
Apr-13	216
May-13	193
Jun-13	230
Jul-13	212
Aug-13	192
Sep-13	162
Oct-13	189
Nov-13	244
Dec-13	209
Jan-14	207
Feb-14	211
Mar-14	210
Apr-14	173
May-14	194
Jun-14	234
Jul-14	156
Aug-14	206
Sep-14	188
Oct-14	162
Nov-14	172
Dec-14	210
Jan-15	205
Feb-15	244
Mar-15	218
Apr-15	182
May-15	206
Jun-15	211
Jul-15	273
Aug-15	248
Sep-15	262
Oct-15	258
Nov-15	233
Dec-15	255
Jan-16	303
Feb-16	282
Mar-16	291
Apr-16	280
May-16	255
Jun-16	312
Jul-16	296
Aug-16	307
Sep-16	281
Oct-16	308
Nov-16	280
Dec-16	345

• Calculate both the three-month and the five-month averages for these data.
• Plot the data to examine the possible existence of trend and seasonality.
• Prepare the following two (2) separate forecasting models to examine the thermostat's sales data using monthly data:
• An exponentialsmoothing model (=0.38, smoothing constant for the level)
• Holt's model (=0.04, smoothing constant for the level; =0.07, smoothing constant for the trend)

d. Examine the accuracy of the forecast given by each model ( four models: MA 3, MA 5, Simple Exponential smoothing and Holt's Method)by calculating the root mean square error (RMSE) for each during the historical period.

e. Which model does minimize the RMSE? Carefully explain which characteristics of the original data caused one of these models to minimize the RMSE.

f. Using Holt's methodforecast 12 months of thermostat sales for 2017.