Category: Uncategorized

I have been thinking about how to engage students with R packages that may, at first, seem to make life incredibly easy for them, but which could make it more difficult for us to find assessment offenses.

The easystats family of packages is one such case.

Getting them to think about additional features is one such case. For example, in the report package having an option for variable names that are more nicely formatted than how they appear in the dataset would allow for the automated text output to be a lot more elegant and publication ready – even just using the classic iris dataset as an example.

For example: see the following example, as generated using quarto.

The ANOVA (formula: Sepal.Length ~ Species) suggests that:

• The main effect of Species is statistically significant and large (F(2, 147) = 119.26, p < .001; Eta2 = 0.62, 95% CI [0.54, 1.00])

Effect sizes were labelled following Field’s (2013) recommendations.

using the lm command

model <- lm(Sepal.Length ~ Species, data = iris)
report.output<-report(model)

The automated reporting (stored in report.output) is as follows:

We fitted a linear model (estimated using OLS) to predict Sepal.Length with Species (formula: Sepal.Length ~ Species). The model explains a statistically significant and substantial proportion of variance (R2 = 0.62, F(2, 147) = 119.26, p < .001, adj. R2 = 0.61). The model’s intercept, corresponding to Species = setosa, is at 5.01 (95% CI [4.86, 5.15], t(147) = 68.76, p < .001). Within this model:

• The effect of Species [versicolor] is statistically significant and positive (beta = 0.93, 95% CI [0.73, 1.13], t(147) = 9.03, p < .001; Std. beta = 1.12, 95% CI [0.88, 1.37])
• The effect of Species [virginica] is statistically significant and positive (beta = 1.58, 95% CI [1.38, 1.79], t(147) = 15.37, p < .001; Std. beta = 1.91, 95% CI [1.66, 2.16])

Standardized parameters were obtained by fitting the model on a standardized version of the dataset. 95% Confidence Intervals (CIs) and p-values were computed using a Wald t-distribution approximation.

Even the option to be able to replace “Sepal.Length” with “Sepal Length” would really improve appearance of the resulting statements, as would being able to save the actual reference used [which was Andy Field’s 2013 textbook Discovering Statistics using IBM SPSS Statistics].

This would allow students to engage more deeply with package contents rather than be passive users. Perhaps an assessment in the form of a portfolio of suggested ideas (and implementations if possible) of features would be an interesting platform.

Even better, would be using peer assessment to help determine which of the suggested ideas would potentially be the most useful at their level of analysis.

A quick post to cover some of the technical issues associated the move to online teaching and marking.

Audio

I use an Audio-Technica 2020USB+ condenser microphone when at my desk.

In the class room, I previously have used a wireless microphone (with a USB receiver) when recording live content in rooms that I moved around a lot in.

Other audio options I have tried:

• headset microphone – not great, records breathing!
• microphone built into earphones, but dangling lower – better at not picking up breathing, but my one is a bit crackly.
• a simple lapel microphone with a direct line in and a long cable – a good, inexpensive option for those looking for something that won’t take up too much space.

Webcam

Using a webcam built into a laptop means that you will have a horrible angle in most situations – think of all the people you see being interviewed from home!

I have a seperate webcam that can be positioned to my eyeline. This webcam also has document camera software (it is the HUE HD Pro USB document camera) which means that I can also use it as a visualiser and share the application with students for live tutorials.

Lighting when using a webcam

I have my desk set up right beside patio doors, so I have excellent natural light. However, there are some things you should think about here:

Natural light is your best lighting friend, so use it where possible. The nearer you can be to a window, the better.

Try to set up your space so it has a neutral but light background, with minimal things in the background to avoid distracting the viewer.

Avoid having big really dark items as the webcam tries to expose for these and then messes up all the rest of the lighting (for example, my curtains need to be out of shot or else everything else is over exposed. This tip was given to me during a video chat with a videographer, so I was able to have personalised recommendations for fine tuning my space.)

Try to avoid overhead light – it can create shine and shadows.

Have a small source of soft light to one side [if you have frosted rather than clear light bulbs it is helpful, if you don’t you could try a layer of tracing paper to create a softer effect – but you will lose some light]. I have a free standing lamp that has both an uplight and a side light. The side light is height adjustable, so I position that to come at the side of my face.

Harsher lights can be directed from the back (back lighting).

Bounce lights off bright surfaces (perhaps even a sheet hung up) to avoid a glare from direct lights.

Marking / Annotating documents

I use a small Wacom device (CTH-480S-ENES) that I have had for years. Note that the photo below has it on an A4 pad for scale. The pen doesn’t have a battery and is very comfortable to use.

The important thing to remember for this one is that you need to configure the device so that the aspect ratio is appropriate or all of your writing will be distorted.  A free solution for quick annotations is (surprisingly) the MS Edge browser.  Using this for quick annotations does not massively increase the size of the documents.

I’m a statistical ambassador – one of many at the Royal Statistical Society (I think that we are now on the third group).  Occasionally we get interesting requests. This one came in relatively recently.  Having more experience with the business end of things at festivals, I decided to volunteer for it.

Last year, I was part of a team that was based in Glastonbury during the festival, looking at the usage of generators on site.  I have also been involved in running juggling / circus festivals for many years.

While I did also do simulation based tests – looking at how long a random walk through an area would need to be, on average, before getting within a specific distance of a friend  (typically we used the limit of 2m) – given that your friends were also wandering around at random.

I looked at different size areas, including some trapezium shaped spaces (which would reflect the shape of a crowd in front of the stage).  These simulations all lead to less conservative estimates than using the negative binomial distribution lead to.  However, using the negative binomial distribution allows for a lot more situations to be investigated very quickly!

The negative binomial distribution is often used to model how many “successes” you observe before your “nth failure” – or vice versa (you just flip the probabilities).  As the team at the BBC were only interested in the duration of time to find the first friend, this simplifies the calculation somewhat. The resulting article is here.

The code / output below comes straight from the html file produced using Rmarkdown. Using Rmarkdown generated html files was a very easy way of sending the code and results back and forward, where people could access the results on pretty much any device!

I used cumulative sums as the negative binomial looks at the probability of finding your friend after seeing exactly that many people, rather than at less than or equal to that many people. The code below is also designed for situations with relatively dense, large crowds, where a person will only know a very small percentage of the total crowd; it is thus not suitable for modelling how long it would take me to find someone I know at a juggling convention, where the crowds are smaller, I know a large percentage of the overall crowd, and the total crowd is much smaller than at music festivals.

In saying that, it didn’t take me long to find people I knew at Glastonbury once I wandered towards the Theatre and Circus Field!

NegBinomialSearch1<-function(p,maxsteps=40000,maxdist=maxsteps/3,crowddensity=2){
  MaxFaces<-maxdist*4*(crowddensity/2)
  Nstrangers<-1:MaxFaces
  ret<-cumsum(p*((1-p)^Nstrangers))
  return(ret)
}

crowd.density<-2
MaxCrowd<-200000
p1<-(1:500)/MaxCrowd # know between 1 and 500 people, out of the total crowd
max.steps<-40000

MF<-floor(max.steps*4*(crowd.density/6)) # how many columns needed in the matrix
PfindingFriend<-matrix(NA,length(p1),MF)

for(i in seq_along(p1)){
  PfindingFriend[i,]<-NegBinomialSearch1(p1[i],maxsteps=max.steps,crowddensity=crowd.density)
}

# P(finding a friend|know x% of population) = .99, what is minimum x
min(p1[apply(PfindingFriend,1,max)>.99])

## [1] 0.00009

# P(finding a friend|know x% of population) = .99, what is minimum z=x*200000
min(p1[apply(PfindingFriend,1,max)>.99])*MaxCrowd

## [1] 18

# P(finding a friend|know x% of population) = .999, what is minimum x
min(p1[apply(PfindingFriend,1,max)>.999])

## [1] 0.000135

# P(finding a friend|know x% of population) = .999, what is minimum z=x*MaxCrowd
min(p1[apply(PfindingFriend,1,max)>.999])*MaxCrowd

## [1] 27

# P(finding a friend|know x% of population) = .8, what is minimum x
min(p1[apply(PfindingFriend,1,max)>.8]) # expect to spot a friend 4 out of 5 days!

## [1] 0.000035

# P(finding a friend|know x% of population) = .8, what is minimum z=x*MaxCrowd
min(p1[apply(PfindingFriend,1,max)>.8])*MaxCrowd # expecting to spot a friend 4 out of 5 days!

## [1] 7

# P(finding a friend|know x% of population) = .5, what is minimum x
min(p1[apply(PfindingFriend,1,max)>.5]) # expect to spot a friend 1 out of 2 days!

## [1] 0.000015

# P(finding a friend|know x% of population) = .5, what is minimum z=x*MaxCrowd
min(p1[apply(PfindingFriend,1,max)>.5])*MaxCrowd # expecting to spot a friend 1 out of 2 days!

## [1] 3

# P(finding a friend|know x% of population) = .25, what is minimum x
min(p1[apply(PfindingFriend,1,max)>.25]) # expect to spot a friend 1 out of 4 days!

## [1] 0.00001

# P(finding a friend|know x% of population) = .5, what is minimum z=x*MaxCrowd
min(p1[apply(PfindingFriend,1,max)>.25])*MaxCrowd # expecting to spot a friend 1 out of 4 days!

## [1] 2

ceiling(MF/2) # distance travelled

## [1] 26667

min(p1[apply(PfindingFriend[,1:ceiling(MF/2)],1,max)>.99])

## [1] 0.000175

min(p1[apply(PfindingFriend[,1:ceiling(MF/2)],1,max)>.99])*MaxCrowd

## [1] 35

min(p1[apply(PfindingFriend[,1:ceiling(MF/2)],1,max)>.999])

## [1] 0.000275

min(p1[apply(PfindingFriend[,1:ceiling(MF/2)],1,max)>.999])*MaxCrowd

## [1] 55

ceiling(MF/4) # distance travelled

## [1] 13334

min(p1[apply(PfindingFriend[,1:ceiling(MF/4)],1,max)>.99])

## [1] 0.00035

min(p1[apply(PfindingFriend[,1:ceiling(MF/4)],1,max)>.99])*MaxCrowd

## [1] 70

min(p1[apply(PfindingFriend[,1:ceiling(MF/4)],1,max)>.999])

## [1] 0.000585

min(p1[apply(PfindingFriend[,1:ceiling(MF/4)],1,max)>.999])*MaxCrowd

## [1] 117

ceiling(MF/5) # distance travelled

## [1] 10667

min(p1[apply(PfindingFriend[,1:ceiling(MF/5)],1,max)>.99])

## [1] 0.00044

min(p1[apply(PfindingFriend[,1:ceiling(MF/5)],1,max)>.99])*MaxCrowd

## [1] 88

min(p1[apply(PfindingFriend[,1:ceiling(MF/5)],1,max)>.999])

## [1] 0.0008

min(p1[apply(PfindingFriend[,1:ceiling(MF/5)],1,max)>.999])*MaxCrowd

## [1] 160

ceiling(MF/8) # distance covered

## [1] 6667

min(p1[apply(PfindingFriend[,1:ceiling(MF/8)],1,max)>.99])

## [1] 0.000705

min(p1[apply(PfindingFriend[,1:ceiling(MF/8)],1,max)>.99])*MaxCrowd

## [1] 141

min(p1[apply(PfindingFriend[,1:ceiling(MF/8)],1,max)>.999])

## Warning in min(p1[apply(PfindingFriend[, 1:ceiling(MF/8)], 1, max) >
## 0.999]): no non-missing arguments to min; returning Inf

## [1] Inf

min(p1[apply(PfindingFriend[,1:ceiling(MF/8)],1,max)>.999])*MaxCrowd

## Warning in min(p1[apply(PfindingFriend[, 1:ceiling(MF/8)], 1, max) >
## 0.999]): no non-missing arguments to min; returning Inf

## [1] Inf

# don't reach the point at which we find a friend 99.9% of the time at this distance!

I was fortunate enough to manage to obtain tickets to Hamilton in London (long ago, I signed up to a mailing list that gave me early access to tickets).  The original preview period was cancelled as the building works on the theatre were not completed; thus the show I attended was changed to a preview.

Firstly, to avoid chance spoilers, I’d like to make some comments about the theatre itself.  It looks beautiful, but the toilets unfortunately show some signs of the rush to complete.  Poorly signposted, no hot water and missing the simple details such as a hook on the door to hold the door into the Ladies open during the interval queue – instead everyone having to do the limbo to get in or out of the toilets isn’t exactly ideal – and something that could be fixed in 10 minutes with a little thought.

The ticketless ticketing system worked smoothly – people at the door checked your email against your identification and card; the another person swiped your card to automatically print out your ticket.  The unfortunate part of this process is that the resulting ticket isn’t much of a souvenir.

Now that you have decided to read on past the boring bits, onto the parts that could be deemed spoilers.  I won’t go into the plot (you can get a pretty good summary of that on Wikipedia), if you aren’t aware of the songs, try this on Spotify.

By coincidence, my ticket was for Dec 18th, 2017; International Migrants Day. There was a definite cheer for the statement “Immigrants, we get the job done!” during Yorktown – but it’s hard to know if that was explicitly in connection or if it is related to other (political) matters.

Now, onto my thoughts about the show.  It was slick enough that I wouldn’t have considered it a preview – the full “Opening Show” is tomorrow, so I suppose they should have all the kinks worked out by now.  The cast were amazing; my particular favourites were Jason Pennycooke (as LaFayette & Thomas Jefferson) and Michael Jibson as a fabulously camp King George!  Pennycooke really handled Guns & Ships (which I was slightly pensive about prior to the show).

Much has been written about the story, music and lyrics (Lin-Manual Miranda has done an amazing job turning this story into a dynamic musical).  I don’t think that I could honestly add more to those. I really enjoyed the songs, that at this stage, are really familiar to me.

I want to concentrate on the other part of the production – the staging, costume design and the choreography.   The main set looked deceptively simple.  About the only thing that changed during the entire show was the position of some stairs that were on wheels; I have to admit that I was slightly distracted by wanting to double check that they remembered to put on the brakes after moving it into the new position. Smaller props, including tables, benches and boxes for standing on were moved on and off the stage by members of the ensemble (and swings). These flowed as part of the show; hat tip to the choreographer(s) for incorporating these transitions smoothly into the show.  The complex part of the staging was the rotating turntable stage – not the entire stage, but parts of it.  The choreography used this to full advantage and it didn’t feeling like a gimmick; which I thought that it might when I first saw it in action.

One thing that did somewhat surprise me is that how small the cast is!

The costume design team really shone for me.  With some of the cast playing multiple named characters, it was really important that the costumes immediately distinguished between multiple characters.  The changes of costumes also served to highly the different stages of Hamilton’s life; however this is pretty standard for costume design, so while it impressed me, it didn’t surprise me that much.  But how they handled the immediate cues Lafayette/Jefferson; Hercules Mulligan/James Madison; John Laurens/Philip Hamilton and Peggy Schuyler/Maria Reynolds.

I really enjoyed it; it was well worth the trip to London… those who have tickets for a show in the new year, get excited, its great!

I took the decision to avoid writing about opinion polls etc in the run-up to the local elections.  Then the snap election was quickly called, meaning that it’s been several months since I put words on the screen here.  I think that the quality of the opinion polls isn’t that bad, but I think that the pollsters are really struggling to capture the changes in who will turn out.  I am sure that there will be a more formal reflection by the polling companies on this over the next few weeks.

In that time, I’ve been trying to buy a house… which hopefully will be completed in July. Dealing with estate agents etc takes a lot of time!  In other news, I attended Bungay Balls Up this year, as it nicely fell just after the exam period at UWE.  This consisted of a week in a field full of buttercups, glorious sunshine, a trip to the seaside, a dip in a river (wild swimming is great if the water quality is up to it); some G&T’s, a pizza night and generally very little juggling with a lot of socialising with old and new friends.

I’ve also been working with our Equality and Diversity team to try to make the production of statistics and graphs for Athena Swan Self-Assessment Teams a lot more efficient.

And to top it off, I’m involved in a project that means that we will be onsite in Glastonbury next week.  Hopefully I’ll have some interesting stuff from that.  UWE are involved in a number of projects that are trying to improve the long-term environmental impact of the festival; specifically relating to waste management and recycling; processing urine into power and also power management for the production areas (with an aim to reduce the amount of diesel being used onsite to power generators). I’ll be there as a team leader supervising a number of students, none of whom are mathematics or statistics students, but there is definitely room for some in future (but obviously not next year).  The current weather forecast is for a dry Glastonbury. cooling off over the next few days.

It was announced yesterday that Swedish epidemiologist Hans Rosling died (from cancer). To give you a taster of his performance skills – here is a snippet of his work for “The Joy of Stats” broadcast on BBC4.

A full version of this is on Vimeo:

The Joy of Stats from Gapminder on Vimeo.

Kate Allen at the Financial Times did an interesting piece about him in 2014 – where he spoke of being famous but having limited impact; this piece also told of his successful sale of the software underpinning many of his charts to google, creating a funding source for the Gapminder foundation.

To my surprise, the announcement of his death and peoples responses to it was trending worldwide on twitter.  Thus, considering my last few posts, I thought that it would be fitting to do a quick analysis of what was being said. Looking at tweets between 2017-02-07 14:35:59 UTC and 2017-02-08 10:20:15 UTC.  This required some extra processing as the number of emojis keeps on increasing, and R does not recognise these naturally –  thus requiring an external list of how R encodes each emoji.  This is still not perfect, thus some manual editing of the wordclouds are still needed to remove emojis that are represented as Japanese characters!

The sentiments expressed in the tweets were largely positive, and, considering the man in question, suitably full of joy and trust/hope rather than sadness.

The first of these represents the number of tweets that expressed each sentiment [note that a single tweet can express multiple sentiments]

While this image represents the strength of the sentiments: