1 / 18

Started in January 1996 by Larry Page, and soon after by Sergey Brin

Background. Started in January 1996 by Larry Page, and soon after by Sergey Brin 2 students at University of CA Originally started on Stratford University website Registered google.com on September 15 th , 1997. What We D id. Search Google for random words of different lengths

blanca
Download Presentation

Started in January 1996 by Larry Page, and soon after by Sergey Brin

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Background • Started in January 1996 by Larry Page, and soon after by Sergey Brin • 2 students at University of CA • Originally started on Stratford University website • Registered google.com on September 15th, 1997

  2. What We Did • Search Google for random words of different lengths • Compare number of letters vs. search time • Compare number of results vs. search time

  3. Procedure • Picked random words from Wordox dictionary • Searched for just that word on Google home page • Copied down number of results and time it took for the search

  4. Assumptions • 2 independent SRS • Actual relationship is linear • Assumed • Assumed

  5. Data

  6. Results Part 1 Pretty scattered Very weak relationship

  7. Line Test Ho: β=0 Ha: β>0 t = b/SEb = -.3576 p(t>-.3576) = .6390 We fail to reject Ho because our p-value of .6390 is greater than alpha = .05. We have sufficient evidence that the slope of the population regression line for search results and time on Google is equal to zero. Therefore, as number of results increases, the time of search stays the same.

  8. Confidence Interval Confidence level: 95% b ± t* SEb (-4 x10-10 , 2 x10-10) We are 95% confident that the slope of the population regression line lies between -4 x10-10 and 2 x10-10 seconds per result. Because zero is within the interval, we can say that there is no relationship between number of results and search time.

  9. Results Part 2 Moderate Strength Positive Linear

  10. Line Test Ho: β=0 Ha: β>0 t = b/SEb = 3.1667 p(t>3.1667) = .00123 We reject Ho for Ha because our p-value of .00123 is less than alpha = .05. We have sufficient evidence that the slope of the population regression line for number of letters and search time on Google is greater than zero. Therefore, as number of letters in the search increases, the search time increases.

  11. Confidence Interval Confidence level: 95% b ± t* SEb (.0032, .0142) We are 95% confident that the slope of the population regression line lies between .0032 and .0142 seconds per character.

  12. Google vs. Yahoo • Compare number of search results on Google and Yahoo • 1 Samp-t test for matched pairs

  13. Data Google results – Yahoo results

  14. Results Part 3 Graph of difference in Google results vs. Yahoo results (Google – Yahoo)

  15. 1-samp t test (matched pairs) Ho: µd = 0 Ha: µd > 0 t = .2806 p(t > .2806|df=29) = .3905 We fail to reject Ho because our p-value of .3905 is greater than alpha = .05. We have sufficient evidence that the mean difference of Google results vs. Yahoo is equal to zero. t =

  16. Sources of error Don’t always get the same search time every time you search for the same word. Internet connection may not be the same during all tests More people searching may affect the search time

  17. Conclusions The number of characters you search for in Google can slightly affect the search time The number of results found on Google may have no effect on the search time Google and Yahoo do not have a significant difference in number of results

More Related