🏫 Schools in Trentino

🎯 Aim of the project

This repository has been created for the Geospatial Analysis and Representation for Data Science course of the University of Trento, teached by Diego Giuliani and Maurizio Napolitano (a.y. 2021/2022).

The aim is to provide some insights about the scholastic system in the province of Trento, in Italy, considering all the schools inside VivoScuola, the official website for schools in Trentino.

💻 The website

The results of this project are presented through a website: https://auroraarctic.github.io/TrentinoSchools, showing:

• All the schools in the province;
• The distribution of schools (private, public and per grade);
• The network of streets around a specific school;
• Some choropleths with population and students' data;
• Points of interest around a school;
• Evolution of Trentino's school network through the KNN algorithm varying the parameter $k$;
• The distribution of students, schools, classes and population among communities.

📁 Repository structure

The repository is structured in three different folders:

• data: it contains all datasets used within the project (population, geographical boundaries, schools and Trentino's data);

• env, with the configuration files to create virtual environments or packages to install in order to execute the code;

• src: jupyter notebooks and Rmarkdown files are contained inside this folder. There is also the html version of the Rmarkdown analysis of schools;

• viz: it is a folder with subfolders containing all the visualizations generated through the code in src folder. There are barplots for points of interest, isochrones, knn images, Moran's I scatterplots, open street map points of interest, treemap and other maps;

• website, with some css and js files to build the website. Only the index.html file is left outside, for Github Pages hosting.

🐍 How to execute the code

Python part

Option 1️⃣: Anaconda Specification File

This option is available for Windows users only with anaconda installed. Inside the env folder there is a file called spec-file.txt, which provides the exact version of packages from conda-forge:

conda create --name geospatial --file env/spec-file.txt
conda activate geospatial

An alternative for all users may be to create a virtual environment through the environment configuration file:

conda env create --file env/env.yaml

In either cases, conda activate geospatial should work.

Option 2️⃣: Pip install

Despite it is highly suggested to choose option 1 through anaconda, still a virtual environment can be created with python and then install the requirements contained inside requirements.txt through the following commands:

# Creation of the virtual environment
python -m venv geospatial

# Env activation on 
# LINUX/ MAC OS based
source geospatial/bin/activate
# WINDOWS based
geospatial\Scripts\activate.bat

# Install requirements
pip install -r env/requirements.txt

Option 3️⃣: Look at Notebooks

Since notebooks are commented and executed with their outputs, you can just read them. This case applies in particular to the R part of the project, whose output is an html file (src/07-SpatialRegression.html).

R Part

For the R part, the installation is pretty easy. You just need to run on Rstudio console the following command:

install.packages(c("geosphere", "tidyverse", "spdep", 
                   "rgdal", "sf", "sp", "rgeos", "terra", 
                   "DT", "knitr", "spatialreg"), 
                 dependencies = TRUE)