Factor maps of a principal component analysis

A choropleth of the first factor of a principal component analysis on 42 African countries (demographic, social and economic indicators). The axis is diverging and centred on zero. FactorLegend adds what such a map needs: the factor and its share of variance, an interpretation text at each pole, a no-data entry and a methodology note.

By default the legend sits below the map. The same map is shown twice: a vertical legend, then a horizontal one. The legend reads high to low: the positive pole (mostly agricultural, young, high fertility) and the negative pole (high GNP and GDP), each marked by an arrow.

from pathlib import Path

import geopandas as gpd

from mappyng import Map, ChoroplethLayer, FactorLegend

countries = gpd.read_file(Path("data") / "afrique_acp_cp1.geojson")

POLES = dict(
    title="Coordinates on the first factor",
    variance=0.588,
    pole_high="Mostly agricultural, under 15, high fertility",
    pole_low="High GNP and GDP",
    note="Standardised PCA on 42 African countries, late 20th century.",
    nodata_label="No data",
    decimals=2,
)


def _factor_map(orientation):
    m = Map(countries, width=720, height="auto", facecolor="#ffffff")
    m.add(ChoroplethLayer(
        countries, column="cp1", cmap="RdYlGn", method="Quantiles",
        num_classes=5, stroke_width=0.2,
        legend=FactorLegend(orientation=orientation, **POLES)))
    m.title("African development", subtitle="First factor of a standardised PCA")
    m.source("Source: ENSG L1 cartostat, TD3 (African country indicators)")
    return m


# Vertical legend, then horizontal: both render on this page.
vertical = _factor_map("vertical")
horizontal = _factor_map("horizontal")

# Hover tooltips. The gallery embeds an interactive SVG from this dict; the
# same call also yields a standalone HTML file.
TOOLTIP = {"columns": ["nom", "cp1"],
           "aliases": {"nom": "Country", "cp1": "Factor 1"}}
interactive = vertical.to_interactive(**TOOLTIP)
# interactive.save("acp.html")