In search of evidence that the situation was temporary I sought out a silver time series. The Perth Mint has a great selection of historic data available for download. They even include exchange rates and the consumer price index (CPI).

The first graph I created was of the London fix monthly average silver price in US dollars. Yes, things are certainly bad, but silver has reached these levels before, around 1980. Certain market characteristics mean that it is possible for a single trader to significantly influence the price of silver. According to Wikipedia, the Hunt Brothers’ silver holdings and futures contracts caused the spike around 1980. Changed trading rules brought the price back down by 1982.

Silver Price (London Fix Monthly Average) USD

While at first glance, the precedence of the price spike gives hope, the current circumstances are very different. The global financial crisis and massive Government debt in many countries is devaluing currencies making precious metals more secure and attractive.

The Australian dollar is closely linked to commodity prices due to the nature of our economy – we are a large exporter of coal, iron ore and wool, so I is hoped the price in Australian dollars would show a rosier (or silvery) story.

Silver Price (Perth Mint Spot Monthly Average) AUD & USD

Other than dampening a few of the spikes, there does not appear to be much difference. The next thing to look at then was the real price of silver, adjusted for inflation. For this I used the Australian all groups CPI (with a base in Jan 1990).

Silver Price (Perth Mint Spot Monthly Average) Real (Jan 1990) AUD

First observation. I am really glad I did not have this hobby in 1980! Between 2005 and 2010 the price remained fairly constant, but the current spike is still visible.

The excel line graph tells a story, but a 3D rendering certainly adds something. And honestly, if I can’t play with the real thing, 3D modelling fills a gap. In the following (and the feature image) each rotation represents 12 monthly spot prices.

Pretty, don’t you think? Next up, a tangible version – digital fabrication – in sterling silver. Perhaps not until the price drops a little though.

Will the price of silver drop to pre 2011 prices? I hope so, but I won’t hold my breath. There are a few European economies to sort out first I fear.

The Data

There are 17 metrics in the Patents dashboard. The unit level of data is applications (and one measure for staff). The application data are sliced different ways to produce the various metrics.

Visual Variables

Position is used in the gauges by a needle showing the magnitude of the metrics.
Colour, it appears, is used indiscriminately.
Size is used in the bar and column charts to show magnitude of the metrics.

Visual Mapping

Gauges are used in BI dashboards as a metaphor for an aircraft dashboard. Usually dashboard gauges include some information about targets or bands, but the gauges in the patents dashboard show only a single number with no indication of whether or not the value is good or bad. The only indication of relative position is the scale of the gauge.  The actual value is repeated on each graphic making considerable detail redundant.

Information Seeking

While is appears the infographic contains extensive information there is only information on 17 metrics. To identify if the agency is performing well the user must read the text. To explore how the agency is performing in a historical context, static time series charts are available my clicking text-based links.

Interaction

This visualization provides very limited user interaction. Standard icons allow the user to print, save and export data.  The graphic prints well, albeit on many pages.  The ‘export data’ option does not appear to work (it returns an empty file). Providing drill down and targets would improve this infographic.
Shneiderman’s ‘visual information seeking mantra’ can still be seen in this visualisation.  The first screen shows an overview.  As the user scrolls more detail is available and a click through provides historical information. It would certainly benefit from more zooming, filtering and details on demand (which may be provided through the data download).

Stephen Few’s version

Few, a well known visualisation minimalist had a go at creating a better dashboard. His intent was to show how little space is needed to present the same information.  He achieved that, but it could be argued at the expensive of aesthetics and interpretability. Neither display does the data justice, although Few’s version is much closer.

Credit

United States Patent and Trademark Office

The Data

The data come from the American Time Use Survey. Neither the data nor the source is linked from the article but a quick search finds the source website with data files back to 2003.
The unit level of the data is the individual respondent and the data are available in numerous formats for download at this level at the source site.
The visualisation shows the data aggregated to various demographic subgroups. Time appears to be recorded in 10-minute intervals, although the tick marks on the graph show 3 hourly blocks. As the data are represented as proportions, the various activities add to 100% at each time slot (an ‘other’ category is provided).

Visual Variables

Position indicates the time of day (x-axis) and proportion of the population (y-axis).
Size of the area shows the proportion of the chosen population undertaking the activity.
Colour is used to distinguish the various activities.

Visual Mapping

A single area chart represents the data. Area charts are appropriate for time series data with a number of categories (in this case, activities). This type of chart is criticised for the low ‘data per pixel/ink’ ratio. It uses considerable ink when printed. Nevertheless, it seems appropriate and effective in this case.

Information Seeking

There is good differentiation in the chosen colours without being jarring. Text labels are only shown for large categories, which is appropriate and reduces clutter. The ability to compare subgroups across times of day would be a useful addition.

Interaction

Interaction is provided by clicking various subgroups in the top left, by moving the mouse across the graph and by clicking on the chart.
Clicking on the chart brings up details of the particular activity for the demographic group selected and provides a comparison of the demographic subgroups. Hovering over the chart brings up additional information in a floating text box.
Shneiderman’s ‘visual information seeking mantra’ is held to a reasonable extent. The main graph provides the overall, a click provides the zoom and selection of a demographic subgroup provides the filter. Further details on demand (for example, demographic comparisons over time) would enhance this visualisation.

Credit

New York Times

Shan Carter, Amanda Cox, Kevin Quealy and Amy Schoenfeld

The Data

Five summary variables are shown as line graphs. Each summary variable (except health) has components that can be viewed in the right pane. The derivation of the five variables is provided in reasonable detail. The source data are not available on the Newsweek site but references are given. While direct access to the data would be preferable, it should be possible to reproduce the scores based on the information provided.
For comparison, the countries can be grouped by income (GNI per capital), population or a selection of geographic and political subsets.
The unit level for the data is country and each country has a score on each of the variables. No country aggregation into regions is possible.

Visual Variables

Position is used to show a country’s score in the y direction and rank in the x direction.
Colour is used to distinguish subsets of population or income, in the ‘snake’ lines and in tabular/text. A red/green split is used when comparing two countries, which, while intuitive would not be apparent to colour blind users.
Value is used on mouse hover to show a countries score on each of the main variables at the same time.

Visual Mapping

The data are represented as line charts, tables and text and a less common ‘snake’ graph. This is appropriate given the quantitative nature of the five derived metrics and the categorical ranks. It would be possible to present the data in a single dimension for each metric but is likely to be harder to interpret.

Information Seeking

The visualisation allows users to explore a countries relative performance in a number of areas. For the most part the limited use of colour enhances the visualisation. The specific details in the right hand pane allow for more exact comparisons.

Interaction

This interactive visualization allows the user to change which countries are compared. The contrast on mouse hover is very low and is a little hard to read but making comparisons does not depend on this feature. Hovering over each point on the chart shows the country’s actual score on that metric, and its overall rank in a floating text box.
Clicking on a point brings up detailed information about the country in the right hand pane and creates a snake graph spanning all of the five line charts (in the vertical direction). A second click creates a second ‘snake’ to allow for comparisons between two countries. A hover over a third point allows for the comparison of a third country.
Reasonable instructions are provided on the various features.

Shneiderman’s ‘visual information seeking mantra’ is held to a limited extent as an overview is provided with more detail available on click and hover. Zooming is facilitated but the selection of country sub-groups.

Credit

Newsweek

Interactive design & programming: Thomas Klepl; Creative lead: Adam Clarkson

Predictably post game conversations have focussed on inaccuracy as a systemic problem in the competition. Some argue that not enough is done to develop the goal kicking skills, particularly encouraging players to have a set routine for their shot at goal.

Before we leap to conclusions let’s see the facts. The chart below shows the average ratio of goals to points since the start of the (VFL) competition. There is a surprising amount of variation in accuracy from year to year but quite clearly, there is an upward trend since the 1950s. In 1956 the average team scored 10 goals 12 behinds. In 2008 it was 14 goals 12.

So there it is. Perhaps all players don’t need a ritual like Matthew Lloyd’s and Fev’s miss was “just footy”!

When it comes to Carlton and West Coast more analysis might be needed. Carlton’s trend is upwards, although not as strong as the competition average. For West Coast there is much less data and possibly reflects the performance over than time. I’ll look at that in another post.

Finally, if you would like to see what the accuracy looks like for each team, here it is.

Data is for the Home and Away Competition to the end of 2008. Acknowledgements: Statistics from the AFL. Analysis undertaken in Tableau Software.

This is the basic dashboard I am using to explain some simple statistics to my 9yo. It kept his attention for more than 2 minutes which is a good outcome around here!

As we are Carlton fans we started with some Carlton match statistics over the last three years. You can download the dashboard as a PDF or as a packaged workbook. If you don’t have the Tableau Software you can download a trial here or a free viewer here.

The interactive version allows you to select teams to compare. In this static image I show Carlton’s performance against non-Victorian teams.

I am playing with other AFL statistics, especially recent wins and attendance stats. If you have any ideas for analysis or would like to see something similar for another team please let me know in the comments. Of course there are a few teams I won’t enjoy too much.