A reminder that when idle queries pop into your head, the answer can often be found using R + online data. And a brief excursion into accessing the Weather Underground.
One interesting aspect of Australian life, even in coastal urban areas like Sydney, is that sometimes it just stops raining. For weeks or months at a time. The realisation hits slowly: at some point you look around at the yellow-brown lawns, ovals and “nature strips” and say “gee, I don’t remember the last time it rained.”
Thankfully in our data-rich world, it’s relatively easy to find out whether the dry spell is really as long as it feels. In Australia, meteorological data is readily available via the Bureau of Meteorology (known as BoM). Another source is the Weather Underground (WU), which has the benefit that there may be data from a personal weather station much closer to you than the BoM stations.
Here’s how you can access WU data using R and see whether your fuzzy recollection is matched by reality.
I keep seeing years represented by coloured bars. First it was that demographic tsunami chart. Then there are examples like the one on the right, which came up in a web search today. I even saw one (whispers) at work today.
I get what they are trying to do – illustrate trends within categories over time – but I don’t think years as coloured bars is the way to go. To me, progression over time suggests that time should be an axis, so as the eye moves along the data from one end to the other, without interruption. What I want to see is categories over time, not time within categories.
So what is the way to go? Let’s ask “what would ggplot2 do?”
Dual y-axes: yes or no? What about if one of them is also reversed, i.e. values increase from the top of the chart to the bottom?
Judging by this StackOverflow question, hydrologists are fond of both of these things. It asks whether
ggplot2 can be used to generate a “rainfall hyetograph and streamflow hydrograph”, which looks like this:
Read the rest…
New Zealand earthquake density 2010 – November 2016
Using R to add data to maps has been pretty straightforward for a few years now
. That said, it seems easier than ever to do things like use map APIs (e.g.
Google, Open Street Map), overlay quite complex data visualisations (e.g.
“heatmap-style” densities) and even generate animations.
A couple of key R packages in this space: ggmap and gganimate. To illustrate, I’ve used data from the recent New Zealand earthquake to generate some static maps and an animation. Here’s the Github repository and a report published at RPubs. Thanks to Florian Teschner for a great ggmap tutorial which got me started.
My own work in bioinformatics to date has not (sadly!) required much analysis of geospatial data but I can see use cases in many areas – environmental microbiology, for example.
The Nobel Prizes. Love them? Hate them? Are they still relevant, meaningful? Go on admit it, you always imagined you would win one day.
Whatever you think of them, the 2015 results are in. What’s more, the good people of the Nobel Foundation offer us free access to data via an API. I’ve published a document over at RPubs, showing some of the ways to access and analyse their data using R. Just to get you started:
u <- "http://api.nobelprize.org/v1/laureate.json"
nobel <- fromJSON(u)
In this post, just the highlights. Click the images for larger versions.
Last week, Mick Watson posted a terrific article on using R to recreate the visualizations in this WSJ article on the impact of vaccination. Someone beat me to the obvious joke.
Someone also beat me to the standard response whenever base R graphics are used.
And despite devoting much of a morning to it, I was beaten to publication of a version using ggplot2.
Why then would I even bother to write this post. Well, because I did things a little differently; diversity of opinion and illustration of alternative approaches are good. And because on the internet, it’s quite acceptable to appropriate great ideas from other people when you lack any inspiration yourself. And because I devoted much of Friday morning to it.
Here then is my “exploration of what Mick did already, only using ggplot2 like Ben did already.”
I’m a big fan of ggplot2. Recently, I ran into a situation which called for a useful feature that I had not used previously: aes_string.
Read the rest…
As so often happens these days, a brief post at FriendFeed got me thinking about data analysis. Entitled “So how many retractions are there every year, anyway?”, the post links to this article at Retraction Watch. It discusses ways to estimate the number of retractions and in particular, a recent article in the Journal of Medical Ethics (subscription only, sorry) which addresses the issue.
As Christina pointed out in a comment at Retraction Watch, there are thousands of scientific journals of which PubMed indexes only a fraction. However, PubMed is relatively easy to analyse using a little Ruby and R. So, here we go…
Read the rest…
…was the tongue-in-cheek title of an image that I posted to Twitpic this week. It shows the usage of the word “novel” in PubMed article titles over time. As someone correctly pointed out at FriendFeed, it needs to be corrected for total publications per year.
It was inspired by a couple of items that caught my attention. First, a question at BioStar with the self-explanatory title Locations of plots of quantities of publicly available biological data. Second, an item at FriendFeed musing on the (over?) use of the word “insight” in scientific publications.
I’m sure that quite recently, I’ve read a letter to a journal which analysed the use of phrases such as “novel insights” in articles over time, but it’s currently eluding my search skills. So here’s my simple roll-your-own approach, using a little Ruby and R.
Read the rest…
GSE and GDS records in GEOmetadb by date
I was reading an old post that describes GEOmetadb, a downloadable database containing metadata from the GEO database. We had a brief discussion in the comments about the growth in GSE records (user-submitted) versus GDS records (curated datasets) over time. Below, some quick and dirty R code to examine the issue, using the Bioconductor GEOmetadb package and ggplot2. Left, the resulting image – click for larger version.
Is the curation effort keeping up with user submissions? A little difficult to say, since GEOmetadb curation seems to have its own issues: (1) why do GDS records stop in 2008? (2) why do GDS (curated) records begin earlier than GSE (submitted) records?
# update database if required using getSQLiteFile()
# connect to database; assumed to be in user $HOME
con <- dbConnect(SQLite(), "~/GEOmetadb.sqlite")
# fetch "last updated" dates for GDS and GSE
gds <- dbGetQuery(con, "select update_date from gds")
gse <- dbGetQuery(con, "select last_update_date from gse")
# cumulative sums by date; no factor variables
gds.count <- as.data.frame(cumsum(table(gds)), stringsAsFactors = F)
gse.count <- as.data.frame(cumsum(table(gse)), stringsAsFactors = F)
# make GDS and GSE data frames comparable
colnames(gds.count) <- "count"
colnames(gse.count) <- "count"
# row names (dates) to real dates
gds.count$date <- as.POSIXct(rownames(gds.count))
gse.count$date <- as.POSIXct(rownames(gse.count))
# add type for plotting
gds.count$type <- "gds"
gse.count$type <- "gse"
# combine GDS and GSE data frames
gds.gse <- rbind(gds.count, gse.count)
# and plot records over time by type
png(filename = "geometadb.png", width = 800, height = 600)
print(ggplot(gds.gse, aes(date,count)) + geom_line(aes(color = type)))