General writing guidelines for reports

Organize into sections:

• Introduction: What is the question you are examining? Why is this question interesting or important? What did you think you would see, going into this study?
• Experimental design: What datasets and methods did you use, what criteria did you use to perform the study, and why did you make these choices? It's important to use the most accurate methods, and using their most accurate settings. If you need to use a less accurate method because of the dataset size, be clear about this -- provide the justification. The important point here is to know the literature well enough to know what the best methods are! (NOTE: you can put the details here of how you ran the study, sufficient detail to make it reproducible, or you can put it at the end of the paper, as a supplementary materials document. See the last bullet for what goes there.)
• Results: This is where you describe what you saw. If you analyzed more than one model condition, give a separate subsection for the results on each model. For each analysis, provide either a figure or a table, and generally figures are better than tables, but make sure to do these well. Refer to the figure or table for each analysis, and explain everything you see. Try to comment on differences between methods, and see if you note any differences between models.
  • For a figure, (a) show error bars (standard error or quartiles), (b) Label axes, (c) Include informative caption that gives all the information necessary to interpret the data, and include the number of replicates that are shown, say if you are showing averages or medians, etc.
  • For a table, there are no axes to provide, but you still need to give an informative caption. Do not give too many decimal places.
• Discussion: summarize what you saw in results. How does the model condition impact the absolute and relative performance of methods? Have you read any other papers where similar or different trends were reported? Here is where you say what was interesting, what was surprising, what met your expectations, etc.
• Conclusions (optional): this is where you can summarize what you learned and perhaps what you would want to do for future work.
• Bibliography: make sure you cited everything you needed to. So, every piece of software you used needs to be cited, plus other things that you say that need publications for justification. Also: it's best if you have read every paper you cite, or at least skimmed it enough to know what is in the cited paper. Finally, every paper in the bibliography should be cited in the paper, and preferably something relevant said about the paper. Do use a journal version (if it's available) instead of a preprint version.
• Supplementary materials: this is where you put any large tables that really don't belong in the main paper. And, if you haven't provided it elsewhere, this is where you put everything about the experimental study so that a reader could do it exactly the same:
  • URLs where the datasets can be found (and if you didn't use all the replicates for a given model condition, which ones you used)
  • every piece of software you used: version number and command (this includes methods that calculate error/accuracy) and the URL for the software

Other

• never copy or even paraphrase text from somewhere else (if you want to use the text, copy it and put it in quotes, and then add a reference)
• everything you say should be either well established and not need a reference, or be supported by data (your own possibly) or a reference to a publication. (Using my textbook as a reference is okay, but then say which chapter or even page provided that information.)
• make sure all methods are compared on exactly the same set of replicates
• whenever possible, use vetted software for basic commands (like computing alignment or tree errors)---do not write your own code for these things, because that would mean you should first publish the code and get it evaluated
• learn to use latex and bibtex
• always helpful to use a spellchecker and grammar checker.
• always helpful to print the document to catch problems.
• check the size of the font in the printed page --- make sure everyone can read it.
• can help if you read it aloud to yourself, and see if you catch any mistakes
• each paragraph should have just one main point, and the first sentence should introduce that point or make that point.
• pay attention to transitions between paragraphs and between sentences.
• grammar checkers and spellcheckers can be very helpful.
• get feedback on your writing from your friends or other students, and seek out the university writing center. Here at UIUC, it's reachable at https://writersworkshop.illinois.edu

Other low-level writing points Read Stanford University's 20 most common errors. But here are some others.

• Don't use terminology without defining it, unless you really expect it to be known by anyone reading the paper (e.g., NP-hard may be fine to use without defining what it means, but "MSC" won't be, and neither will "GTR").
• Be careful about hyphenation. Sometimes it is really needed, sometiems not.
• Bibliographies are often places where there are problems -- missing information, incorrect capitalization, using a preprint instead of a journal version, etc. Look at the printed version to find the errors! Don't assume that copied bibibtex (from Google Scholar) will be correct.
• Be careful with capitalization - proper names require them. Your bibliography may not have capitals where you need them (because of how bibtex works).
• Be careful about commas. Many people have too many, some people have too few.
• Look at the spacing; you should makes sure to have no space at the end of the sentence and before the period.