When loading a dataset (by Choose a file
or Load test dataset
), you have to select the temporal columns, i.e., the columns to be used for the mean growth computation.
Those columns must have a header containing their date, in numeric format (i.e. 1954 and not "POP1954")
As soon as data are loaded, they're displayed in the Base Data
tab, and a table containing the mean growth for each period is also computed. This table can be seen in the Growth
tab, along with it's standard deviation, and a plot representing it's evolution (the polygon around the curve represent +/- 1 standard deviation).
The Simulation
tab then allows to perform some Gibrat simulations, i.e. a computation where, starting from first observed populations, we compute randomly, for each period, a growth rate for each city. Those growth rates are based on a normal distribution around the observed mean and standard deviation.
Computation โ although optimised to run on each core of the server โ can still be long, so, start with a low number of replications (the default 10 for example) and grow according to the time it takes.
For any question, feel free to contact me
This website is released under the GNU AFFERO GENERAL PUBLIC LICENSE v3