Final Data Scultpure 'Rainfall Babies' Documentation

[Andrew Kurniawan]

Rainfall Babies

Data Sets:

Sources:

Rainfall Data - CONCORD GOLF CLUB (066013) 1/1/1999 - 31/12/1999 from the Bureau of Meteorology
Birth Data - RYDE HOSPITAL 1/10/1999 - 30/9/2000 from Ryde Hospital Midwife birth logs

Hypothesis:

My hypothesis for these two data sets was that if you were to take the birth data and set it 9 months ago as conception data, the trend would be on the days there was more rain, there were more fertilisations.

Discovery:

However, on a daily basis this proved to not be the case, and this is justified with a number of reasons. Mainly, not all births are exactly 9 months from the conception day, and secondly fertilisation can occur even a week after the actual 'act'. As a result, unless you were to take each birth case, and find out the exact date of conception it would be unrealistic to expect a matching graph to rain on a daily basis.

Trend/Pattern:

Comparing them by total in a month however, showed that when there was more rain in that month, the number of 'conceptions' increased, and the same trend for less rain. (more information and statistics on this fact in previous Blogs wk 4 presentation, wk 3 presentation.)

Outliers/Inconsistencies:

There are odd months that dont match however, possibly due to some external factors like planned births for certain dates, seasonal births, and conceptions (as a result of a minus 9month date) being included in the wrong months total because they were premature or overdue. Regardless, the trend, although not on a 1:1 ratio relationship, shows significant drops and peaks at the same time the monthly rain data drops and peaks.

Insight/Meaning:

The obvious meaning is that when it rains, people are more likely to stay inside and procreate. On other levels though you can start seeing stories about life and family, and even notice and make comments about society, how many children are actually planned, or on the other hand (a lighter one) how the weather can affect our behaviours and emotions.

Design:

Inspiration/Representation:

I noticed my line graphs and liked how easy it was to identify the relationship between these two data sets using this type of graph. I also wanted to represent the data in a way that related to it, as well as had a nice contained shape, unlike a line graph which just simply keeps going until you decide to stop. I thought a circle and embryo as insipiration, as they are simple shapes with relevance. So converting the time axis of the line graph into a circle would produce a similar shape, almost like a flower or star with many points. I chose a curved line graph to make it feel more organic.

Data Mapping:

Using these circle line graphs I wanted them to be directly comparable by the viewer, so having them ontop of each other i thought would work well. I also wanted the user to explore trying to match the graphs together as the actual birth dates would not fit with rainfall and you would need to adjust it by around 9 months back to find the pattern. And so this resulted in the idea of being able to adjust and turn the graphs by the user themselves.
For the final sculpture monthly totals are used as the two compared graphs, and daily data being used to control the monthly data of each type. The daily data (of the first 4 months of the year) act as keys (literally) to controlling the movement of the rotation of the monthly data, by moving/rotating the monthly data 4 months per 1 revolution of the daily data key.

Process/Final Sculpture:

Early formation:

Before receiving actual data (gathering data from the hospital and BOM took a few weeks), i expected to be using daily data per month to form the circle graphs. So my first idea Circular Puzzle Model was based on that fact and included 24 rotatable rings in total stacked ontop of each other (2 rings per month, rain and birth data, for a whole year). Being critiqued as too simple and shallow, i was insipred by the gear shapes i used to lock in the rings, and explored the use and complexity of gears and clockwork. Also finally receiving the data i would be using, re-enforced and discovered that i would have to use monthly totals in order to show this relationship.

Brainstorming a few ideas, I was left with two that i thought were most approrpiate, one was following an abstract metaphorical and meaning driven path, and the other following complexity and a more 'fun' model path.

- Fitting Puzzle Pieces

This Idea was based on the two data sets fitting inside each other because of the trend, using lots of different data sets as different puzzle peices, only one would truly fit inside the cutout shape (which would be the graph of one of the datasets) and that was the matching data set.

- Moving Gears

Using the fine gear teeth like shape of the daily data to rotate gears that moved in complex ways to effect the other data all simply for aesthetic purposes and for the viewer/user to be able to play with and be amused.

Final Design Idea:

The final idea and sketch was a kind of amalgamation of both previous ideas to join both form and function, aesthetics/amusement with meaning/art. It was simply the two monthly graph rings stacked comparable ontop of each other surround the central gear works. The gears would be controlled by a central gear, which is turned by a special key, and each key would only turn one set of gears which would turn one outer ring. The keys were actually the line graph representation of the daily data.

Iterations:

First design was to fill the inside with many moving gear parts that would just look 'cool'.

I then felt the need to strip it down, to served two reasons, one to unclutter and not include irelevant objects which would just crowd the interpretation of the data visualisation, the other to reduce friction that that many gears would cause. Included as well are working sketches for the display on the key's handle.

Utilising the gears and wanting to make the large graph turn with a relationship to the amount of months the centre key represented, proved to be quite complex and calculated.
At this point, after drawing up the monthly data onto a whole ring, i noticed that 12 values really doesnt produce that much of an interesting shape, making it difficult to compare it to the other data. So I decided to make the graph produce the 12 month values 4 times, so to display the values of that year 4 times on a ring. This produced a lot nicer of a shape, and made it easier to notice the trend. This then also required that 1 revolution of a four-month daily graph key, would turn the outside monthly graph ring one-twelfth of a revolution.
After many calculations these initial plans showed that the exterior monthly graph ring was going to be too large, so i needed to reduce the number of teeth in the outside gear by using double layered gears.

After a few more tweaks and gear teeth number calculations, the model was almost conceptually working, i just had to find a solution to holding the gears together where I couldnt use an axle (for gear rings)

Final Contruction:

The Keys display daily data of 4 months. The labels show which key corresponds to which dataset and what month the circular graph is visualising. The time and amount axis mark out where the graph starts at 0.


The etchings mark out and seperate each month of one year. The heights represent amounts from zero to max of totals each month. Top layer is rainfall, bottom layer is births.



The aim is for the user to matchup the graphs and notice that its around 9 months difference that the graphs align, and therefore is representing conceptions and eventually reach the conclusion of my hypothesis.

Challenges/Difficulties/Problems:

Initially drawing the gears was painfull to do manually in maya, eventually i discovered a Rhino script that let me generate and draw the gears based on parameters i can set myself. With the right settings all the gears would smoothly fit together

Working out the best way to fit the gear sizes together to produce the correct ratio of turns, as well as physically fitting them inside the larger outer ring internal-gear(monthly graph). It almost came down to luck where i was able to produce the right amount of teeth and physical size to all fit together.

Positioning the internal-gears (gears attached to data graphs) couldnt be done simply with an axle because there was no inside, I overcame this by using rails. Physically connecting these pieces however required complete accuracy for it all to work.

The biggest problem occured during the printing process, and is one of the factors that cause the sculpture to function improperly sometimes. The laser cutter was somehow calibrated wrongly and produced lots of heat causing the perspex sheet to actually buckle and bend half way through printing. This caused a lot of the gears and circles that were printed to become slightly oval shapes, which hindered the gear functionality a lot. But in the end it was more about the design, and through some tweaks, filing and greasing, the sculpture seems to function properly most of the time.

I was also aware from the start of my design that spinning the bottom graph would probably cause the the top one to spin as well and vice versa due to gravity and friction. I disregarded this issue because the alternatives, like making only 1 graph spinable, would hinder the meaning of the entire sculpture as well as the relevance of the keys which again influence the meaning negatively.