[FlorianNachtigall]

Summary

• integrate SDG 11.2.1 public transport accessibility into the preprocessing flow
• add public_transport_accessibility to exported city characteristics/percentiles and regenerate public/cities.json
• wire the new metric into UI metadata so it appears in explorer controls and city characteristics
• add one-off mapping notebook archive under preprocessing/public-transport-accessibility/
• keep local merge input CSV untracked via .gitignore

Notes

• this PR does not update README (left intentionally for follow-up)

Map visualization of indicator-city matching results and NaN-fill logic via geographically weighted regression

cities_pt_access_infill_map.html

Closes #6

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[github-actions]

🚀 Preview deployed for this PR:

https://ai4up.github.io/cities-learning-explorer/pr-7/

This comment is kept up to date automatically.

[SimonMontfort]

one specific comment check; otherwise good to go from my point of view.

[SimonMontfort]

country_rename isn't applied here, so 6 countries may get imputed values

This renames the column CTR_MN_NM to country. The values keep the SDG spelling ('Mexico', 'Czech Republic', 'Cape Verde', …), while result['country'] uses the other spelling ('México', 'Czechia', 'Cabo Verde', …).

In gwr_infill_missing, the loop iterates over donor countries and looks up receivers with out[country_col] == country. For 'Mexico' it finds no receivers and continues, and 'México' is never visited. So every missing city in México, Czechia, Cabo Verde, North Macedonia, Palestine and Republic of the Congo silently stays unfilled.

Suggested fix (ideally the country line would go right after loading the shapefile, so it can be reused everywhere):

Suggested change

	"gdf_pt_access['region'] = gdf_pt_access['CTR_MN_NM'].replace(country_rename).map(cities_ipcc.drop_duplicates(subset='country').set_index('country')['region'])\n",
	"gdf_pt_access['population_density'] = gdf_pt_access['Total_POP'] / gdf_pt_access.to_crs(cities_ipcc.crs).area.div(1e6)\n",
	"gdf_pt_access_norm = normalize_popdens(gdf_pt_access.rename(columns={'CTR_MN_NM': 'country'}), popdens_col='population_density',\n",
	"gdf_pt_access['country'] = gdf_pt_access['CTR_MN_NM'].replace(country_rename)\n",
	"gdf_pt_access['region'] = gdf_pt_access['country'].map(cities_ipcc.drop_duplicates(subset='country').set_index('country')['region'])\n",
	"gdf_pt_access['population_density'] = gdf_pt_access['Total_POP'] / gdf_pt_access.to_crs(cities_ipcc.crs).area.div(1e6)\n",
	"gdf_pt_access_norm = normalize_popdens(gdf_pt_access, popdens_col='population_density',\n",

I have not run the entire code though and checked. So before accepting, it may be worthwhile to check this.