jGaboardi
commented
3 years ago Thank you for your suggestion, we extended the Current functionality section of the manuscript to more accurately and thoroughly describe what is present in spaghetti. The basic underlying process for creating a network instance is present (e.g., “2. generate the network representation”), as well as a textual description of secondary functions For example:
- allocating observation point patterns to the network
and
- utilizing observation counts on network segments and network spatial weights within the Moran’s I attribute to analyze global spatial autocorrelation”
Moreover, the code snippets for the two plots demonstrate and visualize the functionality associated with spaghetti, its parent package (libpysal), a required sibling package (esda), and optional dependencies. While the plotting is performed with matplotlib through geopandas, we believe visualization is a powerful tool in science for demonstration and should be retained here. As mentioned above, we decided to include another figure to demonstrate functionality since we are removing Figure 1 (the PySAL Logo) as per #576. This new figure can be found in #576/#606 with the associated caption as follows:
Demonstrating the creation of a network and point pattern from shapefiles, followed by spatial autocorrelation analysis. A shapefile of school locations (blue) is read in and the points are snapped to the nearest network segments (green). A Moran's I statistic of -0.026 indicates near complete spatial randomness, though slightly dispersed.
The "Current functionality" section of the manuscript only shows one of the functionalities of spaghetti, namely the instantiation of a regular lattice network. The rest of the code snippet is more about geopandas and matplotlib. In line with #575 and #580, I think that the section should make a better case for the need of spaghetti in the Python stack (perhaps snapping points?).
https://github.com/openjournals/joss-reviews/issues/2826