Unique peaks check

[Erin-Rooney]

@kaizadp

Here is the fairbanks code for the unique peaks. Can you double check it for correctness?

Thanks!!

Y1_fairbanks/code/fticr-2-nosc.R

Lines 40 to 52 in 0917f5a

	fticr_water_slopepos_unique =
	fticr_data_water_summarized %>%
	# calculate n to see which peaks were unique vs. common
	group_by(formula, cover_type, plot) %>%
	dplyr::mutate(n = n()) %>%
	# n = 1 means unique to CON or FTC Trtmt
	# n = 2 means common to both
	filter(n == 1) %>%
	mutate(uniquepeak = case_when(slopepos == 'Backslope' ~ "backslope",
	slopepos == 'Low Backslope' ~ "low backslope",
	slopepos == 'Footslope' ~ "footslope")) %>%
	left_join(meta_hcoc_water) %>%
	mutate(slopepos = factor (slopepos, levels = c("Backslope", "Low Backslope", "Footslope")))

[kaizadp]

The problem was here:

Y1_fairbanks/code/fticr-2-nosc.R

Lines 20 to 22 in 0917f5a

	fticr_data_water_summarized =
	fticr_water %>%
	distinct(ID, slopepos, cover_type, plot, formula) %>% mutate(presence = 1)

fticr_data_water_summarized represents peaks present within each "treatment", not individual samples. So you don't want to include ID in your grouping.

fticr_data_water_summarized = 
  fticr_water %>% 
  distinct(slopepos, cover_type, plot, formula) %>% mutate(presence = 1)

I tried a Van Krevelen to confirm, and I now get unique peaks across all slope positions.

[kaizadp]

I see some redundancies in your processing code, I'm going to run through everything and add some notes here.

Also, you jump straight into NOSC for your analysis. I would recommend starting out larger scale for an exploratory analysis first. Do Van Krevelens, PCA, etc. before diving into NOSC for unique peaks. You may lose important pieces of the story if you skip the exploratory data analysis.

[Erin-Rooney]

I actually did van krevelen's first but forgot to put that code.

Y1_fairbanks/code/fticr-4-vankrevelin.R

Lines 185 to 211 in 0917f5a

	# this does only unique loss/gain by open vs. canopy
	fticr_water_covertype_unique =
	fticr_data_water_summarized %>%
	# calculate n to see which peaks were unique vs. common
	group_by(formula, slopepos, plot) %>%
	dplyr::mutate(n = n()) %>%
	# n = 1 means unique to CON or FTC Trtmt
	# n = 2 means common to both
	filter(n == 1) %>%
	#PROBLEMS START NOW
	mutate(loss_gain = if_else(cover_type == "Open", "open", "Canopy")) %>%
	left_join(meta_hcoc_water) %>%
	mutate(slopepos = factor (slopepos, levels = c("Backslope", "Low Backslope", "Footslope")))
	fticr_water_covertype_unique_common =
	fticr_data_water_summarized %>%
	# calculate n to see which peaks were unique vs. common
	group_by(formula, slopepos, cover_type) %>%
	dplyr::mutate(n = n()) %>%
	# n = 1 means unique to CON or FTC Trtmt
	# n = 2 means common to both
	# filter(n == 1) %>%
	mutate(loss_gain = case_when(n == 2 ~ "common",
	(n == 1 & cover_type == "Open") ~ "open",
	(n == 1 & cover_type == "Canopy") ~ "canopy")) %>%
	left_join(meta_hcoc_water) %>%
	mutate(slopepos = factor (slopepos, levels = c("Backslope", "Low Backslope", "Footslope")))

I think it's the same issue there:

Y1_fairbanks/code/fticr-4-vankrevelin.R

Lines 21 to 23 in 0917f5a

	fticr_data_water_summarized =
	fticr_water %>%
	distinct(ID, slopepos, cover_type, plot, formula) %>% mutate(presence = 1)

I think I understand the corrections but it would still be good to go over them tomorrow at our meeting.