Ideas for decreasing length

[cwickham]

@erickabsmith Here are some thoughts on decreasing length.

• Make the title take up less space. One option is to redefine the \maketitle command by updating your header to:

  header-includes:
    - \usepackage{setspace}\linespread{2}
    - \usepackage{float}
    - \makeatletter
    - \def\maketitle{%
    - \par\large\textbf{\@title}%
    - \par\normalsize{\@author}%
    - \par{\@date}%
    - \par}
    - \makeatother

  You might also consider shortening the title.

• Resize or rearrange figures.

  • Figure 2, could be shorter and wider, but be careful that font size don't get too small.
  • Figure 3, make shorter by having two columns of panels, maybe split by natural / urban?
  • Figure 4, make shorter by putting these two plots side by side. OA and F1 are directly comparable anyway so aligning the axes isn't so important.
  • Combine Figure 6 & 7. I'm not sure why there are two Landsat images, could you just use one, then put the predictions next to it.

• Rewrite some sections to be more concise

  • 1.1 Background is nice, but lengthy given the total page count you are working with. Reduce in length to get just the main point across, i.e: it's useful to be able to classify land type to be able answer questions about microclimate in cities, but getting high resolution classifications is time consuming. Predicting land type from satellite images is one solution.
  • 1.2 Objective is OK I think. Only place you could be a bit more concise is discussion of "red-star" classes. Can you justify using Hong Kong in just one sentence instead of three. E.g. "Hong Kong has more than 2 polygons of each LCZ".
  • 2.1 Data, one way to reduce this section might be to focus only on what you did (i.e. not what Yoo et al. did). E.g. Where you got your data, how you indentifed polygons, and how you split into test and training. Reduce discussion of Yoo to a sentence along the lines of "slightly differs to Yoo who ... ".
  • 2.2 Random Forests, also feels long given total page count. It's hard to give exact advice here, since all the paragraphs seem necessary just wordy. You might take each in turn and try to make a couple of lines more concise, while keeping meaning.
  • 2.3 Accuracy Assessment don't give the equations for OA_urban and OA_natural, just describe them in a short sentence based on the definition of OA.
  • 3.1 Results See if you can reduce the second paragraph (comparing F1 scores) to a sentence or two, e.g. rather than one sentence for each "grouping", use a colon then list the members and rough F1 range.
  • Sections after this seem pretty good to me.