Analyzing Programming Projects

1
Analyzing Programming Projects

• Stuart Hansen
• University of Wisconsin - Parkside

2
Goals

• Understand what students find engaging and
  frustrating in our programming projects.
• Improve our programming projects for their next
  iteration.

3
The Survey

1. On a scale of 1-10 (with 1 being the lowest and
   10 being the highest), how engaging
   (fun/challenging) did you find this assignment?
2. What was engaging about it?
3. How could it be re-worked to make it even more
   engaging?
4. On a scale of 1 to 10, how frustrating did you
   find this assignment?
5. What did you find most frustrating about the
   assignment?

4
The Survey - continued

• Survey was voluntary and anonymous.
• Administered on date each programming project was
  due.
• 4 semesters of data for CS1 and CS2
• 3 semesters of data of Data Structures and
  Algorithms
• CS0 data from another institution

5
This Presentation

• Analysis of the authors courses, CS2 and Data
  Structures
• CS Program Details
• Small (but high quality) program.
• Class size varied from low of 6 to high of 30.

6
Engagement and Frustration

• Positive correlation in CS 2
• Small negative correlation between engagement and
  frustration means
• No clear explanation for this difference

7
CS 2 Engagement and Frustration
8
Niftiness Defined

• Niftiness Engagement Frustration
• Simple Definition
• Distinguishes nicely among projects with large
  differences in two values.
• Flawed in that it doesnt distinguish among
  projects where engagement and frustration are
  approximately equal (but may be either high or
  low).

9
Niftiness Histogram
10
Worst Assignments

• Three lowest scoring projects
• All were borrowed
• All had instructor related problems
• Bad data set.
• Major writing component that the instructor did
  not adequately discuss.
• Assumed (Java I/O) background that students did
  not have.

11
Best Assignments

• Polynomials (Niftiness 3.93)
• Used (and re-rewritten) since Pascal days
• Steganography (Niftiness 3.44)
• Kenny Hunt SIGCSE 2005
• Rewritten assignment scored much higher than
  first draft
• Huffman Codes (Niftiness 3.33)
• Owen Astrachan, Nifty Assignments 1999
• Word Ladders (Niftiness 3.18)
• Owen Astrachan, Nifty Assignments 2000

12
Engagement/Frustration by Topic
13
Engagement/Frustration by Topic

• Recursion
• 5 or 6 smallish recursive algorithms
• High engagement/high frustration
• Sorting
• The texts all contain the algorithms
• Dynamic Programming
• If asked to develop the algorithm, (no data) but
  it goes off the charts

14
Future Work

• Data left to analyze.
• No discussion here of CS1 or CS0
• Majority of data is from CS1 and CS0
• Correlation between engagement and frustration
  seems to be course dependent.
• Do GUIs make a difference in CS0 and CS1?
• Additional questions in later surveys.

15
Conclusions

• Assignments from SIGCSE work.
• Refining Assignments works.
• Worst assignments were all first timers
• Best Assignments were all old timers
• No evidence that we need to dramatically change
  the discipline.

16
Questions?