I’ve been wanting to play with the data ICTS collects on wireless connections in the library for a while. Dave Lane kindly gathered all the logs for Tuesday the 11th of October from ICTS. We got 7 million lines of logs, and I wrote some code to extract this:
This is pretty rough: it indicates unique wireless devices connecting to library based access points every hour. that means if someone was sitting in level 4 of James Hight, their laptop or phone would be counted once an hour. If they moved up to level 5 at 10:30, it would be counted once on level 4, and then once on level 5. It includes staff and students.
Tuesday the 11th of October was the last week of term, so there was a lot of studying and assignment writing going on!
There is a lot more that could be done with this data, but I think this provides an interesting ‘proof of concept’ of data we already collect, and can question for all sorts of purposes.
That’s a pretty picture Anton! Timeseries would be interesting too. for example, how closely and consistently do these stats measure up to the door counts and head counts, or is there consistent proportionality? Maybe we could use these automatically collected stats as a surrogate?
I have done some maths, and it seems that we might “spend” approximately $4,000 per annum doing head counts in Central Library alone. (I used $20 per hour, but the actual figure will be more or a bit less depending on who is doing the headcount). I have always thought that this time/money could be better spent on other tasks of which we have plenty. If we could fine-tune Anton’s example, we might be able to do this. Headcounts in Central Library alone take approximately 20 minutes as a minimum. Anyone is welcome to check my maths, as I may have made an error, but I think the figure of $4000 is about right. It doesn’t allow for headcounts in our other libraries, but they take much less time, of course.
Our headcount in Central was 550 at 7pm on Tuesday 11 October. The above figures add up to 873 at the same time, excluding Levels 5 and 1. That is a discrepancy of 323.
Some people will have more than one device and that could explain the discrepancy.
Anton, if people move around on the same floor with their mobile devices does that result in the device being detected more than once and therefore being counted more than once? I am thinking here about people walking around with their mobile phones.
These logs could include usernames, so we could be far more exact about numbers of individuals: that information wasn’t given to us for privacy reasons (and fair enough too!) . There are ways around this that respect privacy.
@Caroline: these figures only count one device on one floor once per hour, so moving floors would get you double counted (potentially going up in the lift would get you counted on each floor!) . there is a lot of work to be done, but this is really just a proof of concept showing that this data exists, and what might be extracted from it.
I presume headcount statistics are kept for internal planning/decision-making? They are not currently being used on the LR monthly reports to Alex, nor are they required for reporting to CAUL
I use the headcounts regularly, including twice this year at SMT. They are very useful to highlight the use being made of our libraries at particular times. This impacts on opening hours, staffing, and the perceived relevance of the Library. I know that this does not reflect the value and impact of our services, but numbers matter. I have been telling everyone that there were 1115 people using the Central Library at 3pm on Labour Day. This information is gold and well worth the cost of collection. And at the same time staff are physically present in all the spaces checking in on what is happening.
Which is not to say that the wireless information is not interesting, it is, but counting devices is not the same as counting people. There are too many caveats about the meaning to use it in quite the same way. Anne