CHAPTER 1. METHODOLOGY Google Trends
illustrate the 4 standard regression analysis plots through R. Observe that observation eighteen (July 2005) is definitely an outlier within each piece. We looked into this day and learned that there had been special marketing event throughout July 2005 known as an `employee prices promotion. ' All of us added the dummy variable to manage for this particular observation as well as re-estimated the actual model. The outcomes are proven.
|
log(yt)
|
=
|
2:312 + 0:114
|
¢ log(yt¡1) + 0:709
|
¢ log(yt¡12) + 0:006
|
¢ xt(1)
|
|
log(yt)
|
=
|
2:007 + 0:105
|
¢ log(yt¡1) + 0:737
|
¢ log(yt¡12) + 0:005
|
¢ xt(1) + 0:324 ¢ I(July 2005): (1.4)
|
Each models provide us constant results and also the coe±cients in keeping are comparable. The thirty-two. 4% improve in product sales at This summer 2005 appears to be due towards the employee prices promotion. The coe±cient about the Google Developments variable within (1. 4) means that 1% increase searching volume is related to roughly the 0. 5% improve in product sales.
Does the actual Google Developments data assist with prediction? To solution this query we make a number of one-month forward predictions as well as compute the actual prediction mistake de¯ned within Equation 1. 5. The average from the absolute values from the prediction errors is called the imply absolute mistake (MAE). Each predict uses only the info available as much as the period the forecast is created, which is 1 week into the actual month under consideration.

Comments
Post a Comment