Closed idoroll closed 7 years ago
The "Which parts of the my videos is being watched the most" graph is calculated as follows:
watch_rate = (number_of_raw_views_of_segment) / (number_of_unique_users_who_have_started_the_video)
I think this calculation helps to answer the question: Which part of my videos being watched the most by the people who have started the video?
Again, we can get the "watch rate per student in course" by dividing by the number of students in the course, but I don't think this will change the way the graph looks as it is dividing by a constant.
Add NA as an activity level?? Then everything will be blank. What am I missing?
So, no changes?
On Thu, Jun 29, 2017 at 3:13 PM, AndrewLim1990 notifications@github.com wrote:
- Let's add NA as an activity level
- These changes you've suggested will not affect the appearance of the video graphs. The "Which of my videos is watched the most" graph will only change it by a scale. You are right, the video graphs are constructed via counts, so again, taking these counts and dividing by the same number will not change the way they look.
The "Which parts of the my videos is being watched the most" graph is calculated as follows:
watch_rate = (number_of_raw_views_of_segment) / (number_of_unique_users_who_have_started_the_video)
I think this calculation helps to answer the question: Which part of my videos being watched the most by the people who have started the video?
Again, we can get the "watch rate per student in course" by dividing by the number of students in the course, but I don't think this will change the way the graph looks as it is dividing by a constant.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/AndrewLim1990/mooc_capstone_public/issues/4#issuecomment-312123201, or mute the thread https://github.com/notifications/unsubscribe-auth/AKEQMFmM0uyhGXs7DwgvDCHkAyoZbd2Kks5sJCGkgaJpZM4OJ6bH .
NA as an activity level: A lot of the data for students do not have their activity level recorded on bigquery. The graphs would not be blank. An activity level of NA would just be a way of selecting students where their activity level (number of hours spent) is not known.
So no changes?: A change that I listed above as number 2. can be done. Personally, I can implement these changes for video and general demographics dashboards. I would not feel comfortable altering any others! To be clear, this would purely be a change to the number displayed in the filtering panel. None of the graphs would change (nor should they I think).
I would not add NA, for two different sets of reasons:
Instructor cannot interpret NA. Which students are these? Thus, better not to include them. In this case, <30, 30-5, and >5 may be fewer than overall - but that's fine.
All students with events are logged. Having many NA students can mean two things:
Either way, I would not add N/A.
On Thu, Jun 29, 2017 at 4:21 PM, AndrewLim1990 notifications@github.com wrote:
NA as an activity level: A lot of the data for students do not have their activity level recorded on bigquery. The graphs would not be blank. An activity level of NA would just be a way of selecting students where their activity level (number of hours spent) is not known.
So no changes?: A change that I listed above as number 2. can be done. Personally, I can implement these changes for video and general demographics dashboards. I would not feel comfortable altering any others!
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/AndrewLim1990/mooc_capstone_public/issues/4#issuecomment-312133850, or mute the thread https://github.com/notifications/unsubscribe-auth/AKEQMBRJFBcT_ydQXiHnAgMEb0xsHWN1ks5sJDFdgaJpZM4OJ6bH .
NA has not been included.
Is it still essential to have the same number across the dashboards?
If not, I will be closing this issue!
Andrew
Do we know that NAs are no-activity? Otherwise there may be a bigger problem. I would suggest to triangualte different data sources (match tracklog NA with person-course, for instance).
If there are no issues here, can close it, IMO.
On Mon, Jul 17, 2017 at 3:08 PM, AndrewLim1990 notifications@github.com wrote:
NA has not been included.
Is it still essential to have the same number across the dashboards?
If not, I will be closing this issue!
Andrew
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/AndrewLim1990/mooc_capstone_public/issues/4#issuecomment-315899431, or mute the thread https://github.com/notifications/unsubscribe-auth/AKEQMEtorK2u6jT9YLc4J2QKy1r4w2u9ks5sO9tFgaJpZM4OJ6bH .
I was doing some work on this. The NAs occur when someone registers for the course but doesn't perform any other tracked action. I'm working on removing them in teh SQL scripts. This should save people some headaches.
That is exactly what I hypothesized. Thanks, Matt.
On Mon, Jul 17, 2017 at 10:17 PM, Matthew Emery notifications@github.com wrote:
I was doing some work on this. The NAs occur when someone registers for the course but doesn't perform any other tracked action. I'm working on removing them in teh SQL scripts. This should save people some headaches.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/AndrewLim1990/mooc_capstone_public/issues/4#issuecomment-315960425, or mute the thread https://github.com/notifications/unsubscribe-auth/AKEQMOYSnsD-HKz1y5_yYK0X5hoAfLEJks5sPD_agaJpZM4OJ6bH .
I'm closing this issue because I fixed this while updating the SQL.
Right now the filter values show students active with this type of element. So, if 100 students are in the <30min group, but only 1 of them watch video, the video page will say "1" and not "100" once filtered.
I would like to see this changed, but this is more than the numbers only - this is about how we treat N/A.
Students who do not watch videos should be included in the analysis with a value of 0 rather than not included. If 99 students do not watch videos and one students watches all of them, the average should be 1%, not 100%. Not sure whether this impacts any graphs, as we present counts and not %s, but perhaps. Does it? Also, this naturally applies to all elements, not only videos.
If N/A do not matter to graphs, then either way works for the n/a on the top.