Forum Discussion

Anonymous's avatar
Anonymous
Not applicable
8 years ago

College Class Schedule Analysis

I am trying to figure out how to show time block useage for college course registration data.  This would be most used as well as least used time blocks throughout each day of the week.  The data are: Day, Start_Time, End_Time, CountofID (count of students in time block).  Unfortunately not all time blocks are the same duration, thus lots of overlapping times.

 

I can plot a histogram of frequency by start time but because of the variability in timeblock duration, the plot is not particularlly acurate as to actual in-class or available times.  Ideas?

 

.  

 

 

 

20 Replies

    • Anonymous's avatar
      Anonymous
      Not applicable

      I like it!

       

      We've not fully defined our analysis needs but comparing overlap hours for lower division courses as well as time slot frequency are at the top of our list.  Think “Guide Pathways” and that really encapsulates what we as a community college need to analyze.

    • lanw's avatar
      lanw
      Regular Visitor

      Hi,

      I'm working on a similar task and would like to see how you build it in Power BI.

      Thanks!

      Lan

      • Anonymous's avatar
        Anonymous
        Not applicable

        lanw 

        I ended up taking the section start and end times and creating one row for each minute of the section. I did this by duplicating the columns for start and end times and changing them to a whole number format, then I added a column that extrapolated that by using a custom column with the following formula: {[SSRMEET_BEGIN_TIME]..[SSRMEET_END_TIME]} which creates a row for each value between the times. For example, a 1000 start to 1150 finish would give you all values between 1,000 and 1,150 for that section. You can then filter out any values where the last two characters are greater than 59 (to eliminate 1075, for example, which obviously isn’t a time value).

         

        I also added a time table (see code below) to link the scheduling data to and used that table, along with counts to create a matrix visual with conditional formatting (see photo). I used a hierarchy so you can drill down from hour to half hour, quarter hours and even down to the minute. This allows you to see all of the overlap for the times. Let me know if this make sense!

         

        Code for time table: You’ll need to add this into a blank query

        let CreateTimeTable = () as table =>
         let
         // Similar to our CreateDateTable script, we start with the smallest unit of the dimension, minute
         // There are a fixed number of minutes in a day, so no need for parameters here
         // 525,600 minutes divided by 365 days in a year = 1440 minutes in a day. 
         // Who says we never learn from Broadway musicals?
         MinuteCount = 1440,
         // Now create a Time type list for a total of 1440 minutes, incrementing one minute at a time
         Source = List.Times(#time(0, 0, 0),MinuteCount, #duration(0,0,1,0)),
         // Turn that list into a one column table
         TableFromList = Table.FromList(Source, Splitter.SplitByNothing()), 
         // Change that table's one column to type Time
         ChangedType = Table.TransformColumnTypes(TableFromList,{{"Column1", type time}}),
         // Rename column to Time
         RenamedColumns = Table.RenameColumns(ChangedType,{{"Column1", "Time"}}),
         // Start inserting columns for each unit of time to represent in the dimension
         InsertHour = Table.AddColumn(RenamedColumns, "Hour", each Time.StartOfHour([Time])),
         InsertMinute = Table.AddColumn(InsertHour, "Minute", each Time.Minute([Time])),
         ChangedTypeHour = Table.TransformColumnTypes(InsertMinute,{{"Hour", type time}}),
         // Creating levels in the hierarchy that might be useful for reporting. Omit if not useful to yours
         InsertQuarterHour = Table.AddColumn(ChangedTypeHour, "Quarter Hour", each if [Minute]<15 then [Hour] else if [Minute] < 30 then Value.Add([Hour],#duration(0,0,15, 0)) else if [Minute] < 45 then Value.Add([Hour],#duration(0,0,30, 0)) else Value.Add([Hour],#duration(0,0,45, 0))),
         ChangedTypeQtrHr = Table.TransformColumnTypes(InsertQuarterHour,{{"Quarter Hour", type time}}),
         ReorderedColumns = Table.ReorderColumns(ChangedTypeQtrHr,{"Time", "Hour", "Quarter Hour", "Minute"}),
         InsertHourNumber = Table.AddColumn(ReorderedColumns, "Hour Number", each Time.Hour([Time])),
         NextHour = Table.AddColumn(InsertHourNumber, "Next Hour", each Value.Add([Hour],#duration(0,1,0, 0))),
         NextQuarterHour = Table.AddColumn(NextHour, "Next Quarter Hour", each Value.Add([Quarter Hour],#duration(0,0,15, 0))),
         InsertPeriod = Table.AddColumn(NextQuarterHour, "Period of Day", 
         each if [Hour Number] >= 0 and [Hour Number] < 4 then "After Midnight" else 
         if [Hour Number] >= 4 and [Hour Number] < 8 then "Early Morning" else
         if [Hour Number] >= 8 and [Hour Number] < 12 then "Late Morning" else
         if [Hour Number] >= 12 and [Hour Number] < 16 then "Afternoon" else
         if [Hour Number] >= 16 and [Hour Number] < 20 then "Evening" else "Late Night"),
         InsertPeriodSort = Table.AddColumn(InsertPeriod, "PeriodSort", each 
         if [Hour Number] >= 0 and [Hour Number] < 4 then 0 else 
         if [Hour Number] >= 4 and [Hour Number] < 8 then 1 else
         if [Hour Number] >= 8 and [Hour Number] < 12 then 2 else
         if [Hour Number] >= 12 and [Hour Number] < 16 then 3 else
         if [Hour Number] >= 16 and [Hour Number] < 20 then 4 else 5),
         InsertTimeKey = Table.AddColumn(InsertPeriodSort, "TimeKey", each Time.ToText([Time], "HHmm"), type text)
         in
         InsertTimeKey
        in
         CreateTimeTable

         

  • hi, Anonymous

     

    Consider only END-TIME for your review.
    And use YEAR and MONTH filter for your controls.

     

    Download the visual

    Histogram Chart - Microsoft Corporation

     

    Download sample

    Histogram.1.1.2.0.pbix

     

    Please try and test using it. And please mark the right reply as answer if your issue has been resolved, otherwise, please feel free to ask if you have any other issue.

    Best Regards,
    Rfranca

    • Anonymous's avatar
      Anonymous
      Not applicable

       

      I could be wrong, but I do not think this custom visual will solve the problem I have describe.  The custom histogram visual shows two dimensions of: Bin and frequency (count of occurrences).  If the Bin is defined as the start or end time and then we use count of occurrences for the values, we end up with a visual of time usage but it does not account for the days in the week.  For example, most 3-credit courses meet Monday, Wednesday and Friday.  By not including the day name, we will not see time openings on Tuesday and Thursday because we are looking at the entire year, term or month.

       

      I keep thinking about a calendar type component or heat map visual of day/time but maybe I'm not going about this the right way.

       

      To reintegrate the challenge, I am attempting to identify high and low usage time blocks throughout the week (Monday-Friday).  The data contain everything needed to identify the blocks but how to visually represent this in an intuitive fashion has eluded me.

       

      Thanks!

       

       

       

      • v-jiascu-msft's avatar
        v-jiascu-msft
        Microsoft Employee

        Hi Anonymous,

         

        What's the expected outcome? Maybe you can assign an ID for each usage.

         

        Best Regards,

        Dale

  • Anonymous's avatar
    Anonymous
    Not applicable

    I'm working on trying to solve the same problem in Power BI. Were you ever able to find a solution?

    • Anonymous's avatar
      Anonymous
      Not applicable

      kw127533,

       

      I ended up breaking each day into timeblocks of minutes (1,440 min/day) and then performing a simple comparison of course start time/end time against each timeblock.  This worked for me and I was able to visualize course schedule density by various units of time (i.e. day, week, month, term, year, etc.).  So many of our courses start and end at odd times rather than on the hour so I couldn't think of another way to tackle this challenge other than moving to smaller time units.

       

      I've been sidetracked on other projects but when I get back to it, I want to bring in course geolocation data and overlay that on our campus map to analyse traffic flows and space utilization.

       

      Hope that helps.

       

       

      • Anonymous's avatar
        Anonymous
        Not applicable

        Yeah, I was considering that approach as well but by the quarter or half hour. How did you manage creating the timeblocks? Did you create a column for each minute or use something like R to automate this?