Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
branch: master
Commits on Aug 21, 2014
  1. Re-ran ANOVA scripts to include Effect Sizes (on MBP)

    Jonathan A. Whiteley authored
    H2O
    Cyanobacteria
    ARA (N-fixation)
    Available N
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Oct 27, 2013
  1. Effect Sizes for aov(): test code & TODO

    Jonathan Whiteley authored
    Added test code to summary_ES(), also confirming that this will not work
    (and is unnecessary) for lm, and lme output.
    Updated TODO, marking this task as complete.
  2. Effect Sizes for summary() of aov, aovlist

    Jonathan Whiteley authored
    - Replaced custom summary.aov() function with a summary_ES() function
    that just inserts an 'ES' column to the output from summary()
    - also added wrapper functions for summary.aov() and summary.aovlist(),
    which automatically call summary_ES() after calling the core versions
    from the stats package.
    - see 'jaw.misc.functions.R'
    - Tested with H2O analysis (on a PC)
    - I have not re-run all analyses to update output: much of the ggplot2
    code is deprecated on current versions of R, and I either need to update
    the code, or run it under an earlier version (2.12?)
Commits on Jul 17, 2013
  1. TODO; start investigating calculating Effect Sizes for nested ANOVA

    Jonathan A. Whiteley authored
    Effect Size (ES) for factor = SS_trt / SS_total
    - I just have to figure out how to add that to the summary() output as an additional column (?)
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Dec 15, 2012
  1. Added an Icon (Mac-specific): do not track (.gitignore)

    Jonathan A. Whiteley authored
    Credit to this post for explaining how to make it work:
    http://blog.bitfluent.com/post/173740409/ignoring-icon-in-gitignore
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Dec 14, 2012
  1. Refined figures for thesis

    Jonathan A. Whiteley authored
    - note sure the changes to Fauna graphs are a real improvement
      - the long Y-axis titles are a real pain to fit.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. Moss Growth: Notes regarding borderline significant Fragmentation eff…

    Jonathan A. Whiteley authored
    …ects in t3
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Dec 13, 2012
  1. Moss Growth - Productivity: updated graphs

    Jonathan A. Whiteley authored
    - added clearer panel labels for thesis.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. Tweaks to large multi-panel FxP graphs (H2O, Patch.dwt)

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  3. H2O: fix graph bug, updated graphs

    Jonathan A. Whiteley authored
    - CP4-IO correctly excludes 'intermediate' patches (not "other")
    - added CxP-IO plot (drop 'intermediate' patches from CxP graph)
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Dec 8, 2012
  1. Consistent Y axes for Predator & Grazer plots

    Jonathan A. Whiteley authored
    - added Y.lim1 effect to CFP plot (Chamber x Frag x Position)
    
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Dec 3, 2012
  1. Synth - Partial Regression Plots revised

    Jonathan A. Whiteley authored
    - more sensible axis labels (similar to CH1)
    - consistent Y limits
    - output scale
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Dec 2, 2012
  1. Decomposition plots: points over lines

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Nov 5, 2012
  1. Re-ran ALL analyses (CH1, CH2, CH3, CH4) with new factor level labels…

    Jonathan A. Whiteley authored
    …, and original contrasts.
    
    That was painful figuring out the issue with contrasts.
    The nested ANOVA script changed the settings for contrasts
    	options(contrasts = c("contr.sum", "contr.poly"))
    which is different from the default
    	options(contrasts = c("contr.treatment", "contr.poly"))
    Therefore, running scripts *after* the ANOVA scripts generated different model fits ans `summary()` output (including all the glmulti runs) :(
    I added a line in `lib/init.R` to reset the options to the defaults, and now everything behaves as expected.
    *sigh*
    
    
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. Reset contrast options in `init.R`

    Jonathan A. Whiteley authored
    Nested ANOVA scripts use
    options(contrasts=c("contr.sum","contr.poly"))
    
    But other scripts were written and originally run using the internal defaults
    options(contrasts=c("contr.treatment","contr.poly"))
    
    Ideally, I would reset *all* options during initialization, but I don't know how yet.
    
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Nov 4, 2012
  1. SECC Data template update with new factor levels

    Jonathan A. Whiteley authored
    Forgot to change levels in this file
    Also re-ran to re-generate the template file.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. Re-ran CH4 synthesis analyses

    Jonathan A. Whiteley authored
    coef() and related functions like summary() seem to be returning factor names, but *coded* factor levels (numbers instead of the level labels).  
    This is different behaviour from previous results, and I'm not sure why.  
    I haven't upgraded R, as far as I can remember, so I'm pretty sure that's not it.
    
    If I do figure it out, I will have to re-run everything again, including the glmulti runs :(
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  3. Re-ran CH3 - Moss Productivity / Decomposition analyses

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  4. Re-ran Ch2 - Fauna analyses

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  5. Re-ran CH1 analyses (including glmulti)

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  6. Changed references to 'other' patch positions to "intermediate"

    Jonathan A. Whiteley authored
    Where I could find them
    
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  7. Changed references to "Continuous" to "Contiguous" in code

    Jonathan A. Whiteley authored
    Where I could find them
    
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  8. Recoded 'Continuous' frag to "Contiguous", and 'other' position to "i…

    Jonathan A. Whiteley authored
    …ntermediate"
    
    Have not checked dependencies in code, or tested for errors yet.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  9. Updated figures for CH1 scripts (NFix-cb) for Oecologia (as requested…

    Jonathan A. Whiteley authored
    … by editor).
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Oct 21, 2012
  1. Updated figures for CH1 scripts for Oecologia (as requested by editor).

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Aug 12, 2012
  1. Updated figures (and export settings) to make figure labels bigger in…

    Jonathan A. Whiteley authored
    … relation to plotting area.
    
    For Thesis 1st submission
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. Ecosys-models: made final graphs look "bigger" (larger labels, etc.)

    Jonathan A. Whiteley authored
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Aug 6, 2012
  1. CH4: Decomposition partial regression

    Jonathan A. Whiteley authored
    The sign of the partial regression of Decomposition on Richness changes depending on whether H2o is included.
    If I account for H2O, Richness is negative.
    If I leave H2O out, Richness is positive.
    Either way, the relationship is not significant.
    I interpret this as moisture driving everything, and Richness doesn't explain much of what's left over.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. CH4: Growth - dropped Chamber:H2O interaction for consistency

    Jonathan A. Whiteley authored
    glmulti suggests it's not even important.
    I kept a couple of other interactions out of interest.
    The results don't really change much.
    
    - edits to notes in Cyanobacteria modelling script (will not affect results)
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  3. Decomposition: regression tweaks

    Jonathan A. Whiteley authored
    Reduce # of interaction terms to try to reduce over-fitting.
    Not much effect.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  4. Decomposition: regression modelling first try

    Jonathan A. Whiteley authored
    Results are not saved
    - usual approach, although I think too many interactions are just over-fitting my small sample size.
    - Partial regressions are what I'm most interested in, but I'm not really convinced by them.
      - Richness is coming out as negatively related to Decomposition: it looks more positive to me ...
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Commits on Aug 5, 2012
  1. CH4: remove Nfix prior to partial regression of H2O on Growth

    Jonathan A. Whiteley authored
    The result is exactly the same as the H2O partial regression from Growth without Nfix.
    Now this one script provides all the results I need.
    - I turned off Save.results in alternate Growth-models script.
    - Someday, I might just reverse them.  But this works for now.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  2. CH4: Cyanobacteria regressions without Climate factor

    Jonathan A. Whiteley authored
    - I went to Chamber-only (without Position), to stay consistent with Nfix & Moss Growth regressions.
    - The focus is on the continuous variables, not the experimental treatments anymore: those have been addressed elsewhere.
    
    I kept the modelling with the Climate factor around in a separate script (and glmulti files), but it doesn't save results.
    
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  3. CH4: Separate trial script using 'Climate' pseudo-factor

    Jonathan A. Whiteley authored
    - Using the full Chamber:Position Interaction / 'Climate' takes a lot of explanatory power away from moisture.
      - moisture is confounded in the Chamber:Position interaction to begin with.
    
    I have decided it is better to leave Position out, and just go with Chamber, to leave more for moisture to explain directly.
    There is still the problem of no dry patches in the Ambient treatment, so I will avoid Chamber-interactions to avoid over-fitting and unrealistic predicted values in non-existent dry ambient patches.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
  4. CH4: changed "Cells.m" to "Cells" for regression: Climate pseudo-factor.

    Jonathan A. Whiteley authored
    collapsed all ambient patches to "Ambient", and Chamber Inner & Outer into 'Climate' pseudo-factor.
    
    Signed-off-by: Jonathan A. Whiteley <jonathan.whiteley@mail.mcgill.ca>
Something went wrong with that request. Please try again.