crcresearch / gos Goto Github PK

I successfully installed numpy, pandas, and python 3, but I am having trouble navigating to gosdemo/gos. I was able to use command line to navigate to the GOS folder and can see requirements.txt, but I cant execute steps 2 or 4 on the migration example. I keep getting the error "command not found". The same error comes up when I try to install plotly. Is there something else I need to download?

In the macro migration example:

%cd ~/research/migration/GOS/

and

def file_path(name):
    """
    Shortcut function to get the relative path to the directory
    which contains the data.
    """
    return "./examples/multiscale-migration/data/%s" % name

These will not work on a fresh install of GOS.

It is a good idea to use a consistent style to keep things organized. PEP 8 is pretty standard.

I am working on going through the steps in the GOS read me while using the CRC machine and am coming across errors while executing migration.py and VizDemo.py because pandas will not properly install. I use the command pip3 install numpy pandas and an error message came up saying permission denied: '/opt/crc/p/python/3.6.0/gcc/6.2.0/lib/python3.6/site-packages/pandas'

It seems that some neighboring countries are not being assigned. For example, the border between Sweden and Norway does not show up.

t = neighbors()
t["neighbors"]["Norway"]

nan

t["neighbors"]["Sweden"]

Sweden    Finland
Name: neighbors, dtype: object

Here the border between Sweden and Finland appears in the model, but the border between Sweden and Norway does not.

I looked at the source for the data, Neighbors.csv and is shows the following for Norway and Sweden:

NO,578,Norway,5147790,,nb nn,se,UTC+2 UTC+1
SE,752,Sweden,9723810,FI,sv,en fi,UTC+2 UTC+1

The issue seem to be with the data itself, which is missing some borders.

The README should contain:

• Brief introduction to the directory.
• Documentation on how to install all requirements for running software in the directory.
• Information on how to set up a project using GOS in Jupyter.
• Some nice graphs.

From @johare2:

The brain drain part made it so that the top x% of a country was more likely to move, and this motivation was multiplied by a number inversely proportional to the country's mean income. So the top performing individuals in very poor countries are the most tempted by brain drain.

The example could be a useful resource for showing what GOS does and how it simplifies the process of creating global models; however, the notebook in its current status doesn't accomplish this as well as it could.

Instead of creating the migration scores for each agent at initialization, it makes more sense to calculate these score separately. This would allow the score to be updated as the program iterates through simulations.

This would make importing data easier and likely allow for easier communication with visualization libraries. Instead of using a name for each country and making sure that any new data maps to that name, we can take advantage of the existing standardized codes for each country.

The paper describes each country as having a set of neighborhoods which contain agents. These need to be added to have the model in the program match the paper.

I was able to run the visualization by editing the local code on my computer, but when I commented out the basemap import on the master branch of GitHub and re cloned the files onto the CRC machine, it did not sync with the cloned files on the machine. Is there a better way to edit the code when using the CRC machine?