I have a fascination for data. For the manipulation of data, the extraction of information and just the sheer thrill and challenge of grappling with data. Especially raw data. Looking back on my early studies, this is what I loved about chemistry. It wasn’t so much the chemistry itself (although being on the raw-edge of development is always exciting!), it was the process of obtaining and analysing large sets of data that I really enjoyed. It is no wonder that I ended up returning to studies and focusing on Information Systems.

Now, I am a DBA. Why? Because I believe I have an enormous amount to learn about managing data. I believe a lot of scientists have a *throw-away* mentality towards data and the analysis of data. Experiments should be repeatable, therefore data sets can often be reproduced. Experiments are also, by definition, experimental. So it isn’t surprising to observe scientists running ‘adhoc’ analysis routines and discarding the results or, simply writing them to file – never to be seen again. As a DBA I can appreciate the enormous challenge of managing and maintaining data. I can see a lot of potential for storing data in a model that makes it accessible, usable and potentially informative.

I am increasingly aware that data underpins almost every facet of our lives, both professional and recreational. The *science of data* is now truly ubiquitous, and almost *exclusively* *multidisciplinary* (forgive the pun). Science is no longer performed by scientists alone, it involves programmers, analysts and storage experts. I look back at my years studying and realise the two fields are not disjoint but in fact perfectly compatible. Am I still a scientist at heart? Almost definitely. But not a chemist anymore. I am a DBA, who has a passion for data manipulation and analysis. In the areas of data extraction, transformation and analysis I have a lot yet to learn but am increasingly excited in my growing awareness of this *data-centric *world we live in.

### Like this:

Like Loading...

Hi Nick, Thank you for commenting on my Res blog. Your project looks very challenging – as I am pretty useless for math stuff! Hope we have an opportunity to say hello in the real life.

LikeLike

Hi Yuki – thank you for commenting 🙂 I think the hardest thing about project will be the limited time. Just like you said in your blog, there is so many things that you want to do, but there just isn’t enough time! I am sure we will meet around campus this semester, it is a small IT world there 🙂

LikeLike

Hi; I noticed other of your pages mentioning WBGT calculations. Do you happen to be aware of whether Liljegren’s iterative models for WBGT could be accurately implemented using Solver in Excel, or is that so different a method from the Fortran iterative solution routines he used as to not work correctly? If Solver will work correctly, has anyone actually implemented Liljegren’s model in Excel, for one-case-at-a-time calculation of WBGT? I’ve been trying using only the spreadsheet functions, and the models will reach a Solver solution but so far do not give results matching Liljegren’s (15-20% off). Some of the constants are not in his paper or Fortran program (that I can identify so far), and it is not very clear, for example, if environmental values obtained for solar irradiation are defined the way needed for the model. Any ideas most appreciated.

LikeLike

Hi Lawrence,

Thanks for your comments and questions. Excel’s Solver is completely new to me, having had a little read up on it, I think it would definitely be possible – it would be a lot like taking a set-based approach in a database I would think. To be clear, I have been using Bernard’s model for indoor WBGT (see Kjellstrom, T., Holmer, I., & Lemke, B. (2009). Workplace heat stress, health and productivity–an increasing challenge for low and middle-income countries during climate change. Global Health Action, 2.). For this we assume that wind speed (indoors) is 1 meter per second, which helps to simplify the equation nicely. I imagine the fortran algorithm has some sort of iterative refinement process, we have been using the following algorithm (in psuedocode):

mcpOne and mcpTwo are the values of McPherson’s formula (see Lemke and Kjellstrom, equation 9). The algorithm continues to iterate in an effort to solve for Tw, where McPherson’s formula equals 0.

Again, I have only had a really quick read up about Excel’s Solver but I think it would look a little like this:

TARGET CELL: McPherson’s formula. You are aiming for mcpTwo = 0

CHANGING CELLS: Tw, mcpOne, Ew

you would also have some constants: Td, Tw and Ed

CONSTRAINTS: Continue to change (Tw, mcpOne, Ew) and continue to minimise mcpTwo while Tw < Tw, or until McpTwo crosses y = 0

I tried this exact approach in a database (SQL Server) and it was brilliant! It allowed us to minimise entire sets of inputs really quickly. There are some downsides, in that if you have very low temperatures (well into the negatives) or there is a big difference between Tw and Td, then it requires a lot of iterations. This makes it a slow process when you have very large data sets, but it is quick enough on small data sets.

I can't give my thoughts on your specific questions on Liljegren's model, I have focused on Bernard's model. Solar radiation and the zentih angle seem to be the two most challenging variables to accurately predict. I know Lemke and Kjellstrom have implemented Liljegren's model in a spread sheet using a visual basic macro. Again they made assumptions consistent with indoor conditions, which eliminated the delta(Fnet) / Ah term from Liljegren's model.

If you get it working, I would love to hear how it goes.

LikeLike

Hi Nick. This is very useful.

I have just checked on the ~Kjellstrom, T., Holmer, I., & Lemke, B. (2009). Workplace heat stress, health and productivity–an increasing challenge for low and middle-income countries during climate change. Global Health Action, 2.~ research paper you recommended. There’s only one equations for WBGT indoors there:

WBGTid = 0.7*Tnwb + 0.3*Ta

Are you using it too in correlation with the upper code?

For some reason an incorrect results are obtain if using this WBGTid and upper Tnwb.

Thank you for the reply.

LikeLike

Hi Bernard, not sure if my reply below will notify you. So FYI, please see below 🙂

LikeLike

Hi Bernard,

Thank you for taking the time to write. The algorithm above isn’t entirely complete, and it has a margin of error of +- 0.3 degrees, which isn’t ideal and can lead to small inconsistencies. How far out are your calculations?

I have been playing around with this and made a Shiny App to visualise the relaxation of Tw. The code behind this app is on my GitHub account. In particular, the code file that includes the WBGT calculation is: https://github.com/nickb-/Calculating-WBGT/blob/master/relaxation_Tw/final_code/wbgt.R

You might also like to try this reference: Liljegren J.C., Carhart R.A., Lawday P., Tschopp S., Sharp R.. Modelling the Wet Bulb Globe Temperature Using Standard Meteorological Measurements. The Journal of Occupational and Environmental Hygiene pp. 645 – 655. (2008).

If you’d like, feel free to look me up on linkedin or googleplus – send me a message and we can email or chat.

LikeLike