Lessons from the USS Nimitz #1: Keep it simple — It’s not about the technology

http://www.damniwish.com/2009/06/lessons-from-the-uss-nimitz-2-keep-it-simple—its-not-about-the-technology.html

Check it out.

While as Tushar point out that Excel is not neccessarily the best application for any situation, the beauty of Excel is that:
1. at a pinch it will do a reasonable job of all kinds of crazy stuff;
2. it’s loaded on practically every small business computer;
3. most people are not too scared of starting it up, compared to the more dedicated applications that might do a better job;
4. there’s an excellent support community, because so many people use it.

If MacGyver (a 80′s television hero: basically laid-back, extremely resourceful secret agent who prefers non-violent conflict resolution wherever possible and refuses to carry or use a gun) was a business analyst, I’m pretty sure he’d carry an excel install CD around with him, in lieu of a high-end ‘gun’ program.

I had a thought on how to reduce the data footprint of a country chart, without sacrificing too much resolution where it counts. This is probably needed more with a long and skinny country such as New Zealand than a US state with only 1 coastline. Here’s the concept. Take any three adjacent data points. Run a line from the first to the third point, and calculate how far the 2nd point deviates from that line. If it deviates over a certain limit, leave it. If not, delete it. Reiterate the process untill you get a good balance between data reduction and overall representation. This means where you have a really squiggly coastline with lots of twists and turns, then you’ll have more data points when you really want them.

Multiple iteration of the algorithm could mean some sweeping curves ultimately get turned into straight lines if they sweep at a rate that falls within the acceptable ‘deleletion parameter’ range. But you could have an algorithm that accepts a 2nd arguement to account for this. Perhaps this would do it: Delete that 2nd point if and only if you would not then delete the 3rd point when rerunning the procedure. That is, if the algorithm would fail on points 1, 3, and 4, then don’t run it on points 1, 2, and 3.

It will be fun to watch what happens to a country as you run the algorithm. I’ll have a crack at implementing this when I get a chance.

A dictionary definition from the web (one of many similar) says:
chart –noun
1. a sheet exhibiting information in tabular form.
2. a graphic representation, as by curves, of a dependent variable, as temperature, price, etc.; graph.
3. a map, esp. a hydrographic or marine map.
4. an outline map showing special conditions or facts: a weather chart.

[While the theoretical answer would be yes implementing it would be completely impractical.] – if you can implement it, and it works for what you’re trying to achieve, then why is it impracticle?

[Can you make the colour conditional] – probably…I’ll try to find out. But I might want to look at more than state-wide colors.

[every attempt has reinforced the conclusion that it is the wrong approach.] You state this like it’s a universal truth. If it doesn’t work for you, then you should say ‘for me’.

[Did you visit the page I linked to see what one can do with shapes?] yes. That’s one handy way of presenting the data…one of many.

[Can you click on the state and zoom in to a closer view?] – yes, I believe I can. I’ll whip up an example (hopefully this week) and post a link to it here.

[she is credited with the invention of the polar-area chart!] I’m not a big fan of polar charts myself. I wouldn’t go as far as to say that they are catagorically the wrong approach, though.

Given a XY Scatter chart with a series that looks like the state of Nebraska, can you fill in the state with a color? [While the theoretical answer would be yes implementing it would be completely impractical.]

Can you make that color conditional? [Did you visit the page I linked to see what one can do with shapes?]

Can you click on the state and zoom in to a closer view?

I do things with charts that leave most people scratching their heads in wonder. That doesn’t mean I think a chart is the solution to *every* problem. Quite the contrary. It requires knowing when a chart is not the appropriate tool.

I’ve tried mapping zip codes and postal delivery routes in Excel charts and it is the wrong tool for the task. I’ve tried representing a range of geographic entities and geospatial analysis in Excel charts and every attempt has reinforced the conclusion that it is the wrong approach.

Dr. Snow used a *map* to track the 1854 cholera outbreak, not a chart. How one could use a Excel chart to track the spread of a disease is beyond me. One needs a map. And, a chart is *not* a map.

On the other hand, what Florence Nightingale did with her analysis of causes of deaths in hospitals during the Crimean War would be amenable to graphical analysis. In fact, she is credited with the invention of the polar-area chart!

Essentially, one has to use the appropriate tool for a specific problem. And, a chart is not always the way to go. There are times when Excel is not the way to go. And, then, there are times when neither software, not even computer technology, is the way to go.

For sure, what you’re proposing is all that’s required for business applications that only require data displayed at the aggregated sales area level. But that’s the tip of the iceberg in terms of end use options.

A chart allows you to place specific events right where they occur, so that you can draw conclusions from that. For instance, a modern equivalent of Dr. John Snow (the guy who’s map is credited with discovering the source of of deaths from the 1854 London cholera outbreak) might want to use a user-friendly tool like Excel to track movement of swine flu . Or a resource management organisation might want a friendly interface that helps display very specific environmental indicators.

Given shapes one can do a lot of ‘conditional’ highlighting with them. For an example see
Dashboard example – conditional colors of shapes
http://www.tushar-mehta.com/excel/charts/0301-dashboard-conditional%20shape%20colors.htm

[I am working on a revised version of the page that will fix a few problems with information that is not completely visible on the page and also introduce a simpler way of connecting the shapes to the corresponding values.]

In the above, I copied the individual shapes from one or more websites. Given that one can get map data as you did, one could create a map with VBA using the data and multiple shapes. Then, one could implement the conditional color of shapes.

What would be really great is to have say a macro that adjusts the resolution of points for a particular part of coastline upwards if there’s a good deal of variation (as there would be for a fiord) and downwards if there’s not too much variation (for instance a long sweeping beach) in order to have the most efficient representation of the coastline (i.e. best overall representation for the least amount of dots).

I’ll see what I can come up with (although I’m only up to page 355 of John’s ‘Power programming with VBA’, so my code will be ruder than my previous comment)