FUTURE OF SPREADSHEET PROGRAMMING:A COMPARISON BETWEEN TWO PROMISING TECHNOLOGIES
Author
Zhivkov, Georgi Ivanov
Term
4. term
Education
Publication year
2022
Abstract
This thesis compares two approaches to user-defined functions (UDFs) in spreadsheets: Excel Lambda functions and sheet-defined functions (SDFs) in FunCalc. It addresses three questions: (1) performance, (2) which technology programmers find more familiar and understandable, and (3) whether UDF languages can support teaching Computational Thinking. Performance was evaluated via benchmarks across multiple tasks; sample size was determined using an initial random sample (n=100) and Cochran’s formula. Results were inconclusive overall but tended to favor FunCalc; Excel outperformed on 2 of 7 benchmarks, including a large multi-sheet workbook (Groundwater_daily), where FunCalc’s results aligned with prior work. Future work includes clarifying whether Excel executes in parallel or sequentially and how data representation differences affect speed. Usability was examined through a Cognitive Dimensions analysis and a live-coding user study with participants unfamiliar with the tools; Excel Lambdas were judged more useful, while FunCalc elicited more issues (e.g., use of cell identifiers instead of variable names, lengthy definitions, and the requirement to write functions on a dedicated function sheet). In Excel, participants struggled with parentheses/commas, the Name Manager window, and having the expression directly after the parameters. Educational applicability was discussed philosophically by relating functional programming concepts to UDFs. The thesis contributes an empirical performance comparison, a user-centered assessment of understandability and usefulness, and a discussion of potential for Computational Thinking.
Dette speciale sammenligner to tilgange til brugerdefinerede funktioner (UDF’er) i regneark: Excel Lambda-funktioner og ark-definerede funktioner (SDF’er) i FunCalc. Studiet adresserer tre spørgsmål: (1) ydelse, (2) hvilken teknologi programmører foretrækker og forstår bedst, og (3) om UDF-sprog kan støtte undervisning i Computational Thinking. Ydelsen blev vurderet via benchmarks på tværs af flere opgaver; prøvestørrelsen blev fastlagt med en indledende tilfældig stikprøve (n=100) og Cochran’s formel. Resultaterne var inkonklusive, men pegede mod, at FunCalc generelt var hurtigere; Excel klarede sig dog bedst i 2 af 7 benchmarks, herunder et stort multiarbejdssæt (Groundwater_daily), hvor FunCalcs resultater matchede tidligere fund. Forslag til videre arbejde omfatter at afklare Excels eksekveringsmodel (parallel vs. sekventiel) og betydningen af datarepræsentationer. Brugbarhed blev undersøgt gennem en analyse af Cognitive Dimensions og et brugerstudie med live-kodning blandt deltagere uden forhåndskendskab; her blev Excel Lambdaer vurderet som mere nyttige, mens FunCalc udløste flere problemer (bl.a. celleidentifikatorer frem for variabelnavne, lange definitioner og krav om funktionsark). I Excel gav parenteser/kommaer, Name Manager-vinduet og placeringen af udtryk efter parametre udfordringer. Anvendelsen i uddannelse blev behandlet filosofisk med fokus på ligheder mellem funktionel programmering og UDF’er. Specialets bidrag er en empirisk ydelsessammenligning, en brugercentreret vurdering af forståelighed/nytte og en diskussion af potentialet for Computational Thinking.
[This apstract has been generated with the help of AI directly from the project full text]
Keywords
FunCalc ; Excel ; SDF ; Spreadsheets ; Sheet-defined ; Functional ; Computational Thinking ; Computational ; Cognitive ; Dimensions