MICHAEL TSO

Making a Shiny app with synced leaflet views

2022-01-31T10:01:00+00:00

leaflet.js is a

The leaflet R package by RStudio to allow seamless use of leaflet.js in R codes and Shiny web apps, as I have shown in the PBMS webapp. But how can we help users compare two (or more) maps, especially when they want to zoom in to look at the details.

Two new R pakcages leaflet.mini.charts and manipulateWidget extends the functionality of leaflet to allow syncing of two or more leaflet outputs by simply typing this command:

combineWidgets(mymap1, mymap2, ncol = 2)

| |

You can run a live demo here: https://moisture-leaflet2min.datalabs.ceh.ac.uk/

You can see the code of a minimal example in the gist here:

Write an academic paper in R?

2021-09-14T19:01:00+01:00

R has become a very nice ecosystems that is not limited to programming or analysis but the full web and editing experience. Have you worried that you don't know enough LaTex and Microsoft Word drives you crazy when writing acadmic papers. Well a very nice option is to take advantage of R Markdown and the rticles R package. The latter comes with lots of templates for popular acadmic journals, which essentially turn them into the required LaTex format for you?

What about citations?

You can do @bibkey as you normally would for citation in LaTex using BibTex. Put all your BibTex entries in a .bib file.

Not just LaTex?

R Markdown utilizes 'pandoc' which means it can turn into other document formats such as HTML, Word and others. A cool use case here is the R Journal web document. R Journal doesn't use a publishing house so it compiles the articles themselves. I can simply compile my article in R markdown with the ouput option rjtools::rjournal_web_article. See my example here.

What about track changes?

You can make use of the online latexdiff tool if you have two versions of the R-generated LaTex file (.tex). Then you can just run the diff.tex file with the journal latex template. It's probably easiest to run it on Overleaf. Don't forget to define the latexdiff commands in your LaTex files. There's an latexdiffr package which essentially make this all seamlessly done within R, but I have trouble getting it to work as of today.

Web browser for JupyterLab

2020-06-25T19:01:00+01:00

Part of my job as a data scientist is to develop web apps. We do our work mostly on collaborative cloud platforms. What if I want to preview my app with a local browser? That's quite straightforward in a local machine, but may be tricky on cloud platforms.

Here are a couple ideas of doing in in a cloud-hosted Jupyter Lab notebook.

(1) iframe

JupyterLab support iframes natively. You can try something like as python code:

from IPython.display import IFrame
IFrame('https://statetag-pbmsdemo.datalabs.ceh.ac.uk/', width=700, height=700)

Note that some websites banned being read as iFrames.

(2) JupyterLab externsions

jupyterlab_iframe

This one is allows typing in a web address and open it as an iframe.

jupyterlab_html

This one opens HTML webpages. So you will have to save the HTML file locally.

(3) What about opening a PyQt5 GUI in Jupyter Notebook?

Visualizing bird samples with R Shiny and leaflet

2020-06-24T19:01:00+01:00

leaflet is a very versatile java script library for interactive maps. It is also a great and free alternative to Google Maps, where many of the advanced APIs are not free. Being hooked to R as a package, it is super easy to use and it's my go-to package when I am producing any maps. One application that is of interest is to show the location of the bird samples we have received through the UK Predatory Bird Monitoring Scheme.

At UKCEH, I am working on a R Shiny app that would enable even more customerisation in such interactive maps. This is a quick app that I quickly put together. More development is ongoing to build on this and to give it a lot more analytics capabilities. Stay tuned!

Check out the app here:

Using reticulate: troubleshooting tips

2020-06-08T10:01:00+01:00

reticulate is a R package by RStudio to allow seamless use of python codes in R. Although it works nicely, you may encounter some common issues when setting it up, especially when doing so on a virtual machine.

(1) Sourcing python

The very first thing is to check python is properly installed.

py_discover_config()

If libpython is missing, do this:

$ sudo apt-get install -y r-base python3 python3-dev python3-pip python3-ven python3-tk

(2) Installing packages

use_python("/usr/bin/python3.5") # point this to where your python executable is

You can create a virtual environment to install packages, say it's called "python_environment"

virtualenv_create("python_environment") # create a virtual environment for reticulate

Then you can installed packages

virtualenv_install("python_environment", "matplotlib")

You may also want to check what virtual environments are there, and the path to a particular environment.

virtualenv_list()
virtualenv_python('python_environment')

To use a environment in your code, include use_virtualenv("python_environment",required = T)

(3) Matplotlib

Unfortunately, not all Matplotlib backends work for reticulate. Make sure you put plt.switch_backend('agg') in your python code.

When using this backend, some programs may be missing. You need to install them in the terminal.

$ sudo apt install texlive-fonts-recommended texlive-fonts-extra
$ sudo apt-get install dvipng

https://github.com/rstudio/reticulate/issues/470 https://stackoverflow.com/questions/54477891/filenotfounderror-errno-2-no-such-file-or-directory-latex-latex-python

Checking if everything works

After loading reticulate in R, try running below after typing repl_python(), or as a python code chunk in a R markdown document.

```{python}
use_python("/usr/bin/python3.5") # point this to where your python executable is
use_virtualenv("python_environment",required = T)
import numpy as np
import matplotlib.pyplot as plt
plt.switch_backend('agg')
t = np.arange(0.0,2.0,0.01)
s = 1+np.sin(2*np.pi*t)
plt.plot(t,s)
plt.show()

*Centered text*

1 / 3

Caption Text

2 / 3

Caption Two

3 / 3

Caption Three

❮ ❯