ISSS608 2018-19 T1 Assign Kateryna Mazurenko Conclusion
AIR QUALITY IN BULGARIA
|
|
|
|
|
|
Conclusion Air pollution in Bulgaria shows seasonal trend, it lies within EEA limits in summer and increases multiple times in winter time with peaks in December and January. This seasonal trend most probably is resulted from increased use of energy for house heating and electricity. Solid fuel used for heating and it is one of the major sources of air pollution. Surrounding mountains influence air circulation in Sofia city area causing smogs.
Bulgaria is major exporter of energy to Balkan region so that demand is multiplied with other countries need. Heating season is started officially when daily average temperature becomes 12 degrees or less for three consecutive days.
However, comparing the data with previous periods, situation tends to get better. The highest concentrations of particles were shown back to winter 2013-2014 and never happened again at least for registered period.
Lessons Learned 1. To start with visualisation, we need to pay attention to the data format to be able to build the pipelines and combine all the files using joins. R packages (lubridate) provide a lot of value in preparation for time series.
2. Initial discovery was made using JMP (for example running distributions), I could do it in R as well - as JMP may not be available in future.
3. At the same time, attention has to paid to the data size - for some time, Tableau run out of memory with 10+sheets workbook. I also had to create limited extract to publish dashboard to Tableau public.
4. Overall the design could be improved using stories in Tableau. I decided to come up with buttons for navigation between sheets - to explore new functionality released in the 2018.3 version. The same way, I could do exploration of multiple locations from citizen data to have it combined with time trend in one page, adding value to geographical view. Playing video only gives impression, but not the trend. Colour management in Tableau - I believe data representation benefits a lot with it (diverging colours, consistency etc).
5. Topology data with elevation only was not really useful, as it’s not enough to build contours and/or polygons. Mountains in the North of the city already shown in background.
6. Tableau is a tool which seems to be easy to learn, but I think it’s also hard to master. Simple features like adding reference line for context, filtrations etc provide immediate value - but to dive deeper a lot of additional preparations have to be performed.
