Best approach to using Lat-Long information?

[AndreaNOdell]

What is the best way to start looking at spatial differences using the lat-long information? Currently, the latitude and longitude values for each observation are in separate columns. There are just under 1000 unique latitude and longitude observations each - should I bin the values? Or should I try to visualize them where the x axis is latitude and the y axis is longitude so it is more of a spatial representation?

[AndreaNOdell]

Something like this maybe? If I plot it this way, where 0 (observed value is close to predicted weight at length) is indicated by white points, the whole plot looks pretty white and the more extreme values are hard to see. So this plot actually only shows a subset of the data limited between -0.5 and 0.5 residual values (log10-transformed) and values -0.1 through 0.1 have also been removed to minimize white points. Maybe I could make small-ish bins for long-lat, then summarise the info within the bins to minimize overlapping points. Any thoughts?

[kristinmarshall-NOAA]

Just catching up on responding to these issues and reiterating what we talked about last week here - I like these maps/plots and I think not binning is ok at this point. I think I suggested trying to see if plotting zero values as clear (not white) might help with the overlap. Did that help, or not really?

Other things we could think about would be fitting spatial models to the weight deviates and then plotting those fitted surfaces on maps - that would show up more as a grid or smooth surface than the data alone.

[AndreaNOdell]

Making the points transparent definitely helped with visualizing the points with colors, but it makes me wonder if we're missing some information about areas that tend not to have growth anomalies. I will continue to mess around with it!

Okay great! I will look into some spatial models and we can talk more about it tomorrow (7/7/22)