My Streamlit displays locally but not on Streamlit Cloud.

Deployment
,
1

from pathlib import Path

import json

import pandas as pd

import streamlit as st

import plotly.express as px

import re

import unicodedata

st.set_page_config(page_title=“Accessibilité santé – Bénin”, layout=“wide”)

st.title(“Accessibilité communale aux infrastructures de santé – Bénin”)

# — Chemins robustes (local + cloud) —

APP_DIR = Path(_file_).parent

GEOJSON_PATH = APP_DIR / “communef.geojson”

def norm(s: str) → str:

if s is None:

    return ""

s = str(s).strip().upper()

s = unicodedata.normalize("NFKD", s)

s = "".join(\[c for c in s if not unicodedata.combining(c)\])  # enlève accents

s = s.replace("’", "'")

s = re.sub(r"\\s+", " ", s)

return s

@st.cache_data

def load_geojson(path: Path):

with open(path, "r", encoding="utf-8") as f:

    return json.load(f)

gj = load_geojson(GEOJSON_PATH)

st.caption(f"GeoJSON chargé : {len(gj.get(‘features’, []))} communes")

for f in gj[“features”]:

f\["properties"\]\["commune_id"\] = norm(f\["properties"\].get("Com_norm"))

# — DataFrame depuis les propriétés —

rows = [feat[“properties”] for feat in gj[“features”]]

df = pd.DataFrame(rows)

# — Renommage pour affichage propre —

df = df.rename(columns={

"Com_norm": "Commune",

"hopital_com_csv_population": "Population",

"hopital_com_csv_nb_infra": "Nb_infrastructures",

"hopital_com_csv_niv_acces": "Niveau_accessibilite",

"hopital_com_csv_idx_A_norm": "Indice"

})

# — Nettoyage/formatage —

df[“commune_id”] = df[“Commune”].apply(norm)

df[“Indice”] = pd.to_numeric(df[“Indice”], errors=“coerce”).round(2)

df[“Population”] = pd.to_numeric(df[“Population”], errors=“coerce”)

df[“Nb_infrastructures”] = pd.to_numeric(df[“Nb_infrastructures”], errors=“coerce”)

# Ordre + couleurs imposées

ordre = [“Tres faible”, “Faible”, “Moyen”, “Bon acces”]

df[“Niveau_accessibilite”] = pd.Categorical(df[“Niveau_accessibilite”], categories=ordre, ordered=True)

couleurs = {

"Bon acces": "#006400",    # vert foncé

"Moyen": "#FFD700",        # jaune

"Faible": "#FF8C00",       # orange

"Tres faible": "#B22222"   # rouge

}

st.subheader(“Carte interactive”)

st.sidebar.header(“Filtres”)

sel = st.sidebar.multiselect(“Niveau d’accessibilité”, ordre, default=ordre)

df_map = df[df[“Niveau_accessibilite”].isin(sel)].copy()

fig = px.choropleth_map(

df_map,

geojson=gj,

locations="commune_id",

featureidkey="properties.commune_id",   # champ dans le GeoJSON

color="Niveau_accessibilite",

category_orders={"Niveau_accessibilite": ordre},

color_discrete_map=couleurs,

hover_name="Commune",

hover_data={

    "Population": True,

    "Nb_infrastructures": True,

    "Niveau_accessibilite": True,

    "Indice": True

},



center={"lat": 9.3, "lon": 2.3},

zoom=5.7,

opacity=0.85

)

fig.update_layout(

margin={"r": 0, "t": 0, "l": 0, "b": 0},

height=650

)

st.plotly_chart(fig, use_container_width=True)

with st.expander(“Voir les données”):

st.dataframe(df.sort_values("Indice"), use_container_width=True)
2

Hey there, thanks for sharing your code and welcome to the Streamlit community! :balloon: Your approach is solid and leverages best practices for robust file paths, normalization, and interactive filtering. Here’s a quick summary and a couple of Streamlit-specific pointers:

  • Your code loads a GeoJSON, normalizes commune names, builds a DataFrame, and displays an interactive Plotly choropleth map with sidebar filters and a data table.
  • The use of st.plotly_chart for interactive maps and st.dataframe for tabular data is fully supported and recommended in Streamlit for these use cases.
  • Your color mapping and categorical ordering for accessibility levels are handled correctly with Plotly Express.

If you want to make the map even more interactive (e.g., tooltips, drill-downs), you could also consider using folium with streamlit-folium or pydeck, but for your current needs, Plotly is a great choice.

If you run into any issues (e.g., map not rendering, data not displaying), please share the error message and a minimum reproducible example. Community members are encouraged to jump in with their insights too!

Sources: