English Deutsch Français Italiano Español Português 繁體中文 Bahasa Indonesia Tiếng Việt ภาษาไทย
All categories

If I have a list of presidents and when they were president, I can determine what years we had good presidents. If I have a list of presidents and where they were born, I can determine what locations produced good presidents. But if I combine that data, what good is it? So my question is how could data with both a time and spatial component be of any use?

2007-12-10 11:45:54 · 2 answers · asked by Aaron D 1 in Science & Mathematics Other - Science

This is an example. I don't even feel it is useful. I am trying to figure out how this kind of thinking could be useful. I guess in tracking infectious diseases?

2007-12-10 12:03:38 · update #1

2 answers

In your example, you have discovered the problem of having sparse data containing two uncorrelated random variables. There have been 43 presidents in 230 years, during which time the country has evolved from being essentially a third-world nation to the sole global superpower. In terms of the spatial scales, the geographical extent of the country, and the backgrounds of the men who became presidents, have greatly expanded over that same time period. How would you correlate the success of a Virginia plantation owner who kept slaves as president with the success of the son of a midwest shopkeeper 200 years later? Is there relative success related to the geography or is it the era? There is no way to decouple the two since you don't have multiple samples (e.g., no slave-keeping plantation owners from Illinois were elected president in 1980 (to my knowledge anyway) and Illinois didn't exist as a state in 1780). In short, and I don't mean this as an insult, but your example is horrible in terms of demonstrating the utility of spatio-temporal correlations.

The epidemiology one you allude to is far better, where you can look at the incidence of new cases of a particular disease as a function of time and isolate where it originated. Classic examples of this are tracing HIV in N. America back to the early 70's and the even better example of the London doctor who identified the pathogen responsible for cholera by looking at case clustering around a communal drinking water well.

The bottom line is that spatio-temporal data is of great use provided one would expect a spatio-temporal relationship to begin with. Things that are time-evolving are good places to start in finding such data sets. Presidential quality is not a time-evolving thing, presidents have not gotten better or worse on average since the start of the union. Alternatively, look for spatial correlation and then see if that correlation evolves with time (wave fields on the ocean are good examples of this, where the wave field is correlated over some length and that correlation evolves with time as a function of the wind).

2007-12-11 07:37:29 · answer #1 · answered by gcnp58 7 · 0 0

Time prevents everything from happening at once.
Space prevents everything from happening in one place.
Create a matrix or spreadsheet with time & place axes.
Put your presidents names where date & state intercept.
Then your third dimension can be how well they performed.

2007-12-10 20:01:06 · answer #2 · answered by Robert S 7 · 0 0

fedest.com, questions and answers