[R] Dates as headers causing confusion but needed to convert to Julian days for ANOVA

Mon Oct 25 23:02:58 CEST 2021

Well, both newbies and oldies need to read and follow the Help files
carefully. In this case, note the "check.names" argument of ?read.csv.  You
need to set it to FALSE in your (omitted) read.csv call, because your
strings are not syntactically valid names (follow the "make.names" link to
learn what are valid names). Here is a little example:

> z1 <- read.csv('mydat')
> z2 <- read.csv('mydat', check.names = FALSE)
> z1
  X01.12.2019 X02.15.2020
1           1           5
2           2           6
3           3           7

> z2
  01/12/2019 02/15/2020
1          1          5
2          2          6
3          3          7

However, because your column names are *not* syntactically valid, you'll
have to change them anyway to avoid further infelicities in accessing and
manipulating the data(e.g. see ?names). How you choose to do this is up to
you.

Bert Gunter

"The trouble with having an open mind is that people keep coming along and
sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

On Mon, Oct 25, 2021 at 1:26 PM Philip Monk <prmonk using gmail.com> wrote:

> Hello,
>
> First post - apologies if I get anything wrong - either in describing the
> question (I've only been using R for a week) or etiquette.
>
> I have CSV files of Land Surface Temperature (LST) data derived from
> Landsat 8 data and exported from Google Earth Engine.  I am investigating
> whether the construction of utility-scale solar power plants affects the
> local climate.
>
> I need to tidy the CSV files so that I can use Two-way ANOVA w/repeated
> measures but am having problems due to column headers (necessarily, I
> think) being dates.
>
> Each CSV currently has the following columns:
>
> Buffer
> Values 100-2000 in 100 increments.  Buffers are 100m wide and extend
> outwards from each site boundary.
>
> 24 columns of monthly data.
> Column headers are in date format (currently dd/mm/yyyy in Excel) and
> relate to the date on which the original Landsat 8 image from which the LST
> data are derived was captured.
> I need these dates to calculate the 'Julian day' (1-365.25) for each month,
> and also to extract the Year.
>
> Time
> Currently 1 = pre-construction and 2 = post-construction.
>
> The data frame created when importing one of these CSV's into R looks like
> this:
>
> 'data.frame':   20 obs. of  14 variables:
>  $ Buffer     : int  100 200 300 400 500 600 700 800 900 1000 ...
>  $ X15.01.2010: num  6.09 5.27 4.45 3.39 2.9 ...
>  $ X16.02.2010: num  6.41 5.99 5.61 4.78 4.31 ...
>  $ X20.03.2010: num  8.93 7.38 6.12 5.61 5.61 ...
>  $ X24.04.2011: num  6.28 5.81 5.15 4.54 4.32 ...
>  $ X07.05.2010: num  6.13 5.54 5.35 4.82 4.52 ...
>  $ X08.06.2010: num  7.71 7.4 6.82 6.14 5.82 ...
>  $ X13.07.2011: num  4.07 2.93 2.69 2.47 2.53 ...
>  $ X11.08.2010: num  5.96 5.68 5.38 4.96 4.57 ...
>  $ X12.09.2010: num  5.76 5.15 4.54 3.87 3.46 ...
>  $ X17.10.2011: num  3.16 2.51 2.51 2.06 2.01 ...
>  $ X15.11.2010: num  4.72 3.77 3.24 2.74 2.49 ...
>  $ X01.12.2010: num  4.26 3.516 2.154 1.056 0.315 ...
>  $ Time       : int  1 1 1 1 1 1 1 1 1 1 ...
>
>
> Importing a CSV into R that has a date as a column header (in whatever
> format) causes problems!  R adds the 'X', and converts the separator.
>
> I was using 'gather' and 'pivot_longer' (see below) but the date issue has
> wrecked that approach.  I've tried reformating the date, trying to remove
> the X, and going away to learn more about data frames, dplyr, and readr.
> I'm not making any progress, though, and I'm just getting more confused.
>
> Helped requested
> ~~~~~~~~~~~~~~
>
> How should I proceed to tidy the data such that I can:
>
> *) extract the year and Julian day for each date, then convert the date to
> the name of the month?
> *) create a tidy table with columns for Buffer, Month, Year, Julian day,
> LST (the values), and Time (1 = pre-construction, 2 = post-construction of
> a solar farm).
>
> Prior to deciding I needed to calculate the Julian day for use in ANOVA I
> was doing this (with month names rather than dates - please remember I'm a
> newbie!):
>
> data <- read.csv(...
> attach(data)
> # data_long <- data %>% pivot_longer(!Buffer, names_to = "month", values_to
> = "LST")
> # data_long <- data %>% pivot_longer(!Buffer, names_to = c("month",
> "Time"), names_sep = 13, values_to = "LST")
> data_long <- gather(data, Month, LST, January:December, factor_key=TRUE)
> data_long$Time <- as.factor(data$Time)
> str(data_long)
>
> 'pivot_longer' didn't work, but 'gather' did to create the long data needed
> for ANOVA.
>
> For example:
>
> 'data.frame': 480 obs. of  4 variables:
>  $ Buffer: int  100 200 300 400 500 600 700 800 900 1000 ...
>  $ Time  : Factor w/ 2 levels "1","2": 1 1 1 1 1 1 1 1 1 1 ...
>  $ Month : Factor w/ 12 levels "January","February",..: 1 1 1 1 1 1 1 1 1 1
> ...
>  $ LST   : num  NA 0.803 0.803 1.044 0.475 ...
>
> Suggestions/hints/solutions would be most welcome.  :)
>
> Thanks for your time,
>
> Philip
>
> Part-time PhD Student (Environmental Science)
> Lancaster University, UK.
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

	[[alternative HTML version deleted]]