Closed rvprasad closed 8 years ago
@rvprasad : like the new proposed title. candidate for discussion, for sure
Review by @barik (copied from #78)
Visualizing data is important, or Visualize often and double check
https://github.com/ds4se/chapters/blob/master/stephan_diehl/visualization.md
Visualizing data is an integral part of the research process, and can be used in many different stages of the research process.
The chapter has some jargon (though it's probably obvious for people in information visualization!), and I would try to add short layman descriptions for some of these terms as they are introduced. For example:
The section "Some examples?" also seems to pivot around a single story. Perhaps this section should instead be named something like "Leveraging Visualizations: An Example with Software Repository Histories". Then, have a brief sentence on what a version control history is, followed by the problem: "Data extracted from software repositories may have been corrupted [...]." Although, I'm not sure corrupted is the right word. Maybe "problematic" instead?
Near the end of the article, there is the wording, "If you are not sure, ask a visualization expert." Presumably, if they had an expert handy, they wouldn't need the book! So perhaps give a more concrete set of pointers here. One example might be: "If you are not sure, check literature on information visualization, exploratory data analysis, or the infoviz mailing lists at [...]." I know for me, one of the difficult things in getting started is having the right keywords.
I would end the article on a positive note. Right now it ends with a criticism, which is fine, but the very last sentence of sentences I think should somehow how the reader benefits from the article. Perhaps tie-back to the original theme of how visualization is powerful and should be integral in the analysis process.
Finally, a minor nitpick, but "e.g." and "i.e." get used a lot in the article. To make the language flow more easily, maybe use phrasing like "that is", or "for example" explicitly to add variety.
The chapter length is adequate. I think the section currently called "Some examples?" could use some paragraph breaks. Currently it is a bit intimidating as it is one long paragraph.
The story gets a bit muddled when it introduces "interaction." That was not something I was expecting. So if interaction is important, perhaps it can be intertwined in the example throughout. Otherwise, this might be superfluous and beyond the scope of this sort article.
When I read the introduction, I was thinking that the mantra is something like "Visualizing end-to-end" or "Visualizing throughout the research process" or something like that. I think novices (myself included) tend to think of visualization as the final publication end-product. But this article makes it clear that visualization is not just camera-copy; it's something that should be done often as part of the analysis process itself to gain insight. The title should somehow reflect that, and maybe that can be the story of the article as a whole.
The use of terms like sequence rule mining, parallel coordinates, decision trees, and so on make it clear that there is a lot of richness in visualization above and beyond, say, "bar graphs." I think it's important to keep this, but I would make these terms more accessible by adding a simple explanation for each as you introduce the term.
A few comments
@sdeal Please take a look at the reviews and prepare a new version of your paper by January 13. The reviewers offer great advice on how to make the chapter stronger.
@sdeal Thanks for making the changes. They look good to me.
I'm not sure if I understand the title correctly. Do you intend to say that Visualizations have to be often double-checked? If yes, how about:
Visualizations: It Pays to Double Check
Thanks, Tom
What about "Visualize to explore and validate" or "Make visualization part of your analysis process" ?
I like "Make visualization part of your analysis process" more because it doesn’t restrict the purpose of visualization to exploration and validation.
--Tom
From: sdeal [mailto:notifications@github.com] Sent: Monday, February 8, 2016 6:37 AM To: ds4se/chapters chapters@noreply.github.com Cc: Tom Zimmermann tzimmer@microsoft.com Subject: Re: [chapters] ./stephan_diehl/visualization.md (#98)
What about "Visualize to explore and validate" or "Make visualization part of your analysis process" ?
— Reply to this email directly or view it on GitHubhttps://github.com/ds4se/chapters/issues/98#issuecomment-181396130.
@tzimmermsr @sdeal
what do i tell the publishers? is this good to go?
Good to go. I've updated the title and status in the Perspectives Book Status.
Title of chapter
Visualizing data is important
URL to the chapter
https://github.com/ds4se/chapters/blob/master/stephan_diehl/visualization.md
Message?
Visualization can serve as a verification lens during data analysis
Accessible?
Yes, the chapter is accessible to the general audience.
An image of decision trees and parallel coordinates can help emphasis "visualization technique can make a big difference" and provide additional info about decision trees and parallel coordinates. Similarly, providing a sequential rule example might help make the exposition more self-contained, e.g., fread should be preceded by fopen.
Casting the pitfalls as headings of paragraphs (e.g., Seeing is not believing, Using off-the-shelf tools, Scalability) in the last section can highlight the message/takeaways.
Size?
The chapter is about the right length.
It might help to add examples to make the pitfalls "stick" more. Say, an example of wrong/right kind of visualization to use with specific sort of data can help the message about choosing the right visualization technique.
Also, since it is a chapter about visualization, it might help to add visuals to reinforce the message, e.g., hair-ball graph vs parallel edge splatting.
Since the last paragraph is a call-to-action to data analysts using visualization, it might help to hoist it as a section.
Gotta Mantra?
How about Verify via Visualization?
Best Points
Don't change the use of decision trees and parallel coordinates in the first section.