Skip to main content
10 events
when toggle format what by license comment
Jan 17, 2018 at 20:55 comment added Matthew Drury You need to think about what is random in your problem. Logistic regression assumes that the response is random, but governed by a probability. In your case, the boundaries seem random, while the observations between the boundaries are deterministic.
Jan 17, 2018 at 20:48 comment added Nick Cox As said, you seemingly have two data points each trip, arrival and departure time. The rest of the data is just ornamental extra.
Jan 17, 2018 at 19:43 comment added dfqupwndguerrillamailde @NickCox what approach do you have in mind?
Jan 17, 2018 at 19:40 comment added Nick Cox I still think that approach is not promising. It is a backward approach to modelling this kind of data.
Jan 17, 2018 at 19:38 comment added dfqupwndguerrillamailde @NickCox I could add days into the data so add data for 7 days of the week as the bay will be used differently on weekdays than it is used on weekends? More data means more variety hence a better prediction...?
Jan 17, 2018 at 19:30 comment added Nick Cox Broadly, my view is that it's unsuitable for your data for the reasons stated. In real data you could presumably focus on times of arrival and departure, which could be expected to vary.
Jan 17, 2018 at 19:25 comment added dfqupwndguerrillamailde @NickCox Thanks for the in depthanswer. It actually helped me understand it more. I came up with this data as I am using it to predict when a parking bay is occupied. These are pretty much dummy values. In the above data, person A parks his car until 9amish and parks it back at 6pmish. I was under the assumption that logistic regression would be suitable for this approach as there are only 2 outcomes; 0 or 1.. what do you reckon?
Jan 17, 2018 at 19:07 comment added Matthew Drury This was almost exactly the answer I was writing : )
Jan 17, 2018 at 18:46 history edited Nick Cox CC BY-SA 3.0
added 70 characters in body
Jan 17, 2018 at 18:39 history answered Nick Cox CC BY-SA 3.0