Talk:Lesson02

The pre-lesson readings highlighted the common pitfall which many people commit during presentations – not thinking through properly on how best to communicate their ideas/findings across to the audience. It is a pity that the hard work of data acquisition, processing and analysis would be eroded by such ineffective dissemination.

The authors flagged up by Dr Kam have interesting comments on the poor use of charts and plots. I particularly like the articles by Stephen Few (in case you are not able to borrow the book from library, there is a set of presentation slides online, with script that comes with a good dose of sarcasm). Links below for sharing please.

- Show Me the Numbers Designing Tables & Graphs to Enlighten, 2004 (https://www.courses.washington.edu%2Finfo424%2F2007%2Freadings%2FShow_Me_the_Numbers_v2.pdf&usg=AFQjCNHLYeHtj323sJLbvurF2ySDOBsavw&sig2=biX7V3Rpn0lK6f52b6TopA)

- Eenie, Meenie, Minie, Moe: Selecting the Right Graph for Your Message, Sep 2004 (https://www.perceptualedge.com/articles/ie/the_right_graph.pdf)

- Are mosaic plots worthwhile, Jan/Feb/Mar 2014 (https://www.perceptualedge.com/articles/visual_business_intelligence/are_mosaic_plots_worthwhile.pdf)

Cheers!!

---

Designing a good visual chart or graphical representation is an art, coupled with the science of Visual Psychology. The science in designing the chart governs the basic mistakes to avoid such as improper use of colours, meaningless graphic, highlighting non-critical part of the chart, etc. The art involves beautifully arranging the info to make the chart easily readable and understood. Equally important is transforming the data into the “correct” format, based on the type of data in order to extract meaningful insights.

Raymond.goh.2015 (talk) 11:52, 23 August 2016 (SGT)Raymond Goh

---

There is no absolutely wrong or absolutely 'right' chart. However, in most presentation, creating chart is to facilitate the communication not to confuse them. We all as analysts have to bear this in mind when creating charts. As mentioned by Stephen Few in this article, it makes no senses to use line graph for discrete/categorical data. Line graphs convey 'movement' and is suitable for time series or other continuous variables. Another part is my personal preference, if it's categorical data, I'd prefer to display it in horizontal bar graph with values as X-axis. I found that this better reflect the fact that there is no relationship between each of the categories, especially in the monthly report where all data are time series.

The figure showed how the same data can be portrayed into line, vertical bar, and horizontal bar chart.

Kanokkorn Prasongthanakit (Lin)

---

This session highlighted the common pitfalls in data visualisation, of which some is familiar to me as I have committed those 'crimes' before. Like what Lin has mentioned, there is no fixed solution in data visualisation. It all depends on the target audience and message one wants to convey to the target audience. Because of that, data visualisation can be misleading (and biased) as it frames the thoughts of readers on certain topics, which may be inaccurately represented on the charts. This was what happened in the Ritz Carlton and Marriott example mentioned in class, where the choice of pie charts to display hotel occupancy hid the real, underlying pattern present in the data. Ignoring such insights could cause negative repercussions to the business. As such, it is also important to explore data thoroughly and choose the right charts to accentuate the trends/patterns the data contains.

Serene.su.2015 (talk) 22:21, 26 August 2016 (SGT)Serene Su

---

This week's topic on the principles and best practices to design good graphs and charts drives home the point that each chart created needs a lot of care and attention in order not to provide a misleading picture to the readers of the charts. Charts that show key messages immediately to the readers are what we should strive for. Some key pointers I have learnt in the readings to apply and practice:

1. Know who are your audience, who are you creating the charts for. Understand their level of understanding charts and what decisions they need to make via the charts (from the article "The Dataviz Design Process")
2. Reduce Clutter - if you don't need a reason for the ink spent, likely you don't need it in the chart. (from the article "The Dataviz Design Process")
3. Guideline to selecting colours - limit to a palette of 2 or 3 colours.Make the hues distinct, or chroma variations of a hue to differentiate items in a chart. (from the article "Choosing Colors for Data Visualization")
4. Using line charts to connect unequal time intervals are misleading, because doing so affects the readers' perception of the trend of the data. (from the article "Line Graphs and Irregular Intervals: An Incompatible Partnership")
5. Normalise numbers to make comparisons easier. Find ways to adjust large numbers (e.g. GNP in trillions of dollars) into numbers where people "on the ground" can appreciate (from the article "Best Practices for Understanding Quantitative Data")

Yongjian.2015 (talk) 03:27, 28 August 2016 (SGT) Chia Yong Jian