In class this week we learned about logistic regression and I will be using my Sephora dataset to practice on so lets begin.
I will be using this model:
My output from R is below:
As we can see here, the predictors along with the intercept are all significant which is represented by the three stars (***) listed after the p values. Since all are significant we can build a model equation.
This is the model equation:
log(p/p-1) = .664 – 0.0000023*Loves + 0.0463*Difference in Price – 0.194*Rating
Where p is the probability that there is a marketing flag listed on a product
The intercept and slopes for Difference in price and ratings are all economically significant where as the slope for loves is not economically significant.
The intercept of .664 can be interpreted this way: for products with a low rating, a low difference in price and a low amount of loves it is likely that there could be a marketing flag on that product. The intercept is slightly higher than 0 so we have a slightly greater than 50% chance of “success” when all X variables are equal to 0. *Remember the rule of thumb for interpretation; when log odds ratio < -3 then p is near 0, when log odds ratio > 3 then p is near 1, and when log odds ratio = 0 then p is .5.
For Loves, we can say that increased amount of loves decreases the likelihood that the product has a marketing flag.
For Difference in price, we can say that an increased amount in difference in price increases the likelihood that the product has a marketing flag.
For Ratings, we can say that an increased rating decreases the likelihood of the product having a marketing flag.
Even though the slopes vary greatly, we cannot say for sure which variable has a greater influence on likelihood of the product having a marketing flag. This is because all of the variables have different scales for example: ratings only go on a scale from 0-5 where has loves can range from 0 to 5 billion.
Thank you for following along!!
NicoleMISY262 