Calculation of relevance of a product?

Asked

Viewed 83 times

-1

Let’s say I have the following data:

Video A:
    Procurado: 12000 vezes
    Clicado:   7000 vezes

Video B:
    Procurado: 11000 vezes
    Clicado:   10000 vezes
  1. How can I calculate the relevance of these two videos in my website?
  2. Should I base myself on other numbers than these? If so, which ones?
logic

  • This data seems to me illogical to apply any relevant standard. If it were for example, Video B searched 10000 times and clicked 4000 times and Video A searched 7000 times and clicked 6000 times, I think it would be more coherent

  • Not always the person would need to search for an item to click on it, it may be that they have clicked on the item through a link coming from another page, for example a suggestion on the home page

  • Because it is, in this case the suggestion of the home page should be counted as Wanted, so that one can make a calculation based on statistical relevance, so that the number of the Clicked is always a percentage of the Wanted

  • I changed the data, as such a calculation could be made?

  • I’ll do the calculations and post an example as an answer

1 answer

1


Video A Wanted 12000 Clicked 7000 - percentage A = 58.33% = 0.58

Video B Wanted 11000 Clicked 10000 - percentage B = 90.9% = 0.90

Calculation of the relevance between Video B in relation to video A

Proportion of the pooled sample (p) - Combination of the proportions of the two groups.

      7000 + 10000
p = _________________ = 0,739
      12000 + 11000

Standard error - Measure the difference between its proportion and the true proportion. A smaller number means that the proportion is closer to the true proportion. A higher number means that the proportion is far from the true proportion

SE = ((0,739*(1-0,739))*(1/12000 + 1/11000))1/2

  SE= 0,00578

Test statistic (t) - A statistic-t. The amount of standard deviations between the distance of a number and the average. t= (percentagem B - percentagem A)/SE

t= (0.9 - 0.58)/0.00578 = 55,36

Statistical relevance - If the absolute value of the test statistic is greater than 1,96* times the standard deviation of the mean, it is considered a statistically relevant difference.

*1.96 is the number used for the 95% confidence level. See page 10 of this link Statica Aplicada

Completion As the absolute value of the test statistic 55,36 is much larger than 1.96, which means that the difference between Video B and Video A is highly relevant.

Browser other questions tagged logic

You are not signed in. Login or sign up in order to post.