logseq/pages/hls__Selecting Third-party Libraries The Data Scientist’s Perspective.md

file-path:: file://C:\Users\david\Zotero/storage/2BBTJQTC/Selecting Third-party Libraries The Data Scientist’s Perspective.pdf

• . ls-type:: annotation hl-page:: 12 id:: 63161a5c-232b-4b5e-a697-1a458ccac2ed
• We asked the focus group participants about what these comments meant to them and provide these potential interpretations in the #paper rather than excluding the two factor comments(See third paragraph of “Additional Factors mentioned by Participants” in Section4.2). hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 ls-type:: annotation ls-type:: annotation hl-page:: 12 ls-type:: annotation hl-page:: 12 id:: 63161a6c-294d-4329-baf3-15f92d365c97
• including 34 who indicate that there are no additional factors that they consider. ls-type:: annotation hl-page:: 31 id:: 63161a95-85b5-45e5-a9c5-a353b900c23b
• we are able to see that data scientists put much value on community support and are constantly looking for easy guidance that can reduce the time they need to learn a new library. ls-type:: annotation hl-page:: 31 id:: 63161aa9-2040-42e6-a8ca-8187dbfc970a
• better adaptability and integration ls-type:: annotation hl-page:: 31 id:: 63161ab0-7631-4a68-b3b7-939be6c0f604
• e use the focus group results to understand data scientists’ library selection tooling needs, which emphasizes the motivation of our work. Overall, adding the focus groups addresses various comments by of all three reviewers ls-type:: annotation hl-page:: 3 id:: 63161ae3-9128-4006-be73-6085b94fa0e1
• 3.3 Focus Groups ls-type:: annotation hl-page:: 25 id:: 63161af8-d2a9-4ce5-8b94-16f7742657e2
• Therefore, we use the focus group method, which has been advocated as a quick and effective way to collect qualitative feedback [50], to gather additional insights to help interpret our survey results. hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 id:: 63161b17-a97e-4895-9518-c8d641ab978a
• We design the focus groups as semi-structured group conversations where we discuss our survey results and interpretations with data science experts hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 id:: 63161ba4-316c-4937-946b-892bb75fe3c1
• It is worth noting that we design the focus group questions and recruitment criteria based on the results of the survey since our goal is to use the focus groups to help us with interpretation and to avoid any researcher speculation. hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 ls-type:: annotation ls-type:: annotation hl-page:: 25 ls-type:: annotation hl-page:: 25 id:: 63161bdc-e056-4b24-a433-f9785bb5dd37
• We conduct each focus group as a semi-structured, unbiased conversation for60-75 minutes addressing the following main points of discussion ls-type:: annotation hl-page:: 26 id:: 63161c23-faee-47bd-8b54-0c4aab8049c5
• From the above, we conclude that data scientists draw our attention to the core role that statistics play in their work. As such, statistical soundness is a factor that may influence their library selection (R19 and R54). Moreover, the nature of the data scientist’s work comes with some concerns about consistency and scalablity (R32, R46, R414) and customization (R71). All of these additional factors can be categorized as technical ones. ls-type:: annotation hl-page:: 33 id:: 63161c49-c77f-4c40-bf7a-d431793c88b7
• RQ2: We find the following statistically significant differences between the two populations. When compared to software developers, 7 factors influence data scientists’ decision more (how much a #library fits the desired purpose, the final customers of the project, the project/product managers, #type of industry, cultures and policies, the community experience, and how active the community is) while 4 factors have less influence over data scientists’ decision (#library security, its active maintenance, maturity and stability, and its license). hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 ls-type:: annotation ls-type:: annotation hl-page:: 36 ls-type:: annotation hl-page:: 36 id:: 63161c54-6e30-43f5-9802-ee84fe1d4249
• RQ1: We find 7 important factors for data scientists: 5 are technical (usability, fit for purpose, documentation, maturity and stability and performance),2 are human (activeness and experience), while none are economical. Additionally, data scientists do consider three other factors when selecting libraries: statistical soundness, consistency and scalablity, and customization. hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 ls-type:: annotation ls-type:: annotation hl-page:: 33 ls-type:: annotation hl-page:: 33 id:: 63161c5b-4934-4f2f-afba-6b8b57925ab2