100202 CS838 Vis - topic: "WHY we might do visualisations" spectrum: Know the question <-----> Don't know the question - what is the right statistics for your question? => Tukey: "exploratory data analysis" ...get to know your data so that you can choose your statistics. "...start with obvious ones, draw lots until something pops out" scagnostics (scatterplot diagnostics) - Tufte: "Visualisation supports reasoning" - Ware chapter => dance metaphor: back and forth interaction, guiding partnership to move things along. => Implications: support pattern finding optimize cognitive process account for economics of cognition attention - Tufte chapter e.g. Snow's case. - utilized 'grouping' methods - showing the empty spaces as data as well - issues of outliers (would be ignored as noise in other fields) ...but it may be outliers for only one kind of visualisation! => How about other visualisations? (e.g. scatterplot) - issue of aggregation: implementing sampling theories kernel density modification, blurring, smooth surfacing, aliasing... ...some you can do, some you shouldn't - Basic: scan, summarize, select a subset => the results shown can differ by different summarization of the visualisation