Mixing writing and `julia`

Author

Denis Schluppeck

Session date:

2023-01-04

Background

Here is an example of a document that produces a plot from some data that’s stored separately.

The data in ?@fig-timeseries shows the daily interactions with the moodle page for my second year lab classes. Can you spot the two dominant patterns in the data?

an actual computed figure

#| label: fig-timeseries
#| fig-cap: "A line plot of some mystery data"

# julia script
#
# schluppeck, 2022-12-12

using CSV
using Dates
using DataFrames
using Plots
using FindPeaks1D

gr()

df = DataFrame(CSV.File("2017-mysteryTimeseries.csv"))

the_plot = plot(df.theTime_day, df.n;
         lw = 2, 
         xlims = Date.(("2016-09-01", "2016-12-31")),
         legend = false
)
ylabel!("# of moodle interactions")
xlabel!("day")

or a table

A badly formatted table… rstats with various packages handles tabular data much more nicely!

# get head()... first 10 rows
head(d) = first(d, 10)
head(df)

or some further analysis

If you want to compute things for including in your text, so-called inline code, then you can make your code spit out markdown that’s been patched up. If you turn #| echo: true to false, then the code is hidden!

We can also elaborate on previous plots, bu adding additional analysis. ?@fig-timeseries-with-peaks shows the days, where \(n>100\).

#| echo: false
#| label: fig-timeseries-with-peaks
#| fig-cap: "Events with > 100 interactions labelled"

# use findpeaks to find days where 
# interactions topped 100
p = findpeaks1d(df.n, height = 100)

# and superimpose on previous plot
plot(the_plot, size = (600,300), widen = 1.1)
scatter!(df[p[1],:theTime_day], p[2]["peak_heights"]) #, color=:red, ms=12)

Notes

to find out which kernels you can use with jupyter, you can check in terminal with

jupyter kernelspec list

Check out how conveniently the output format can be swapped out with quarto render 01-doc-with-julia.qmd --to pdf etc