Your data analysis solution
Name is required.
Email address is required.
Invalid email address
Answer is required.
Exceeding max length of 5KB

Which statistical model should I choose to process exported data from Google AdWords?

By Michal Spisak | Oct 17, 2016 11:08PM CEST

Hi,
I recently use trial version of your awesome! addon, and I asked on the Google AdWords forum here https://www.en.advertisercommunity.com/t5/AdWords-Tracking-and-Reporting/Which-statistical-model-should-I-choose-to-process-exported-data/td-p/844375

Seems, no one knows how to process the data from AdWords.

Your addon is full of options. Could you please navigate me, how to proces user behavior, I am sure you know how AdWords exports looks like, there are lots of data, for example date, city, search query, cpc, impressions, location, region..

I think there is BIG POTENTIAL for every AdWords user to use your addon to estimate future user behavior, future trends.

Do you have some tips I can share on AdWords forum for others what one can do with your addon? Myself, I am currently lost.
I am university student, we had 2 semesters statistics, but not realy in that deep.

Thank you so much!
Best regards
Michael

Best Answer
By Jean Paul | Oct 18, 2016 07:09PM CEST | XLSTAT Agent

Dear Michael,

Thanks for your message. Using statistical models to gain insight into AdWords data could be very interesting. We are not really aware of what’s usually done outside the Google Analytics simple metrics and summary statistics. Or even if something additional is done.
Many things could be thought of:
1) Using log-linear regression to model count data on a certain period of time. For example we could imagine a data set with keywords in rows, a Y variable with number of converted clics and many X variables.
2) Doing the same thing but using ads in rows instead of keywords.
3) Using linear regression to analyze the effect of several X variables on the quality score using the same table as in 1.
4) Using ARIMA to forecast time series
5) All of the above models can be considered with row weights corresponding to the inverse of your bids (this would more or less give the same weight to each keyword / ad).
Well… If you have more ideas to share, let us know!

Cheers


Up

0
By Michal Spisak | Oct 18, 2016 04:19PM CEST

anyone here please?

Up

0

Down

By Jean Paul | Oct 18, 2016 07:09PM CEST | XLSTAT Agent

Dear Michael,

Thanks for your message. Using statistical models to gain insight into AdWords data could be very interesting. We are not really aware of what’s usually done outside the Google Analytics simple metrics and summary statistics. Or even if something additional is done.
Many things could be thought of:
1) Using log-linear regression to model count data on a certain period of time. For example we could imagine a data set with keywords in rows, a Y variable with number of converted clics and many X variables.
2) Doing the same thing but using ads in rows instead of keywords.
3) Using linear regression to analyze the effect of several X variables on the quality score using the same table as in 1.
4) Using ARIMA to forecast time series
5) All of the above models can be considered with row weights corresponding to the inverse of your bids (this would more or less give the same weight to each keyword / ad).
Well… If you have more ideas to share, let us know!

Cheers

Up

0

Down

By Jean Paul | Oct 25, 2016 04:57PM CEST | XLSTAT Agent

Hello Michal,

What do you think about all of this?

This question has received the maximum number of answers.

Contact Us

1c26995d494fb3061dd0ae8571ffc0a4@xlstat.desk-mail.com
https://cdn.desk.com/
false
desk
Loading
seconds ago
a minute ago
minutes ago
an hour ago
hours ago
a day ago
days ago
about
false
Invalid characters found
/customer/portal/articles/autocomplete
9283