Overview
skillviz is an R package for skill co-occurrence analysis and visualization from Online Job Advertising (OJA) data. It provides tools for computing skill specialization indices, co-occurrence networks, source stability filtering, salary analysis, professional distance clustering, and temporal skill ranking. The package is built on ESCO and CP2021 (Italian professional classification) taxonomies.
Installation
Install the development version from GitHub:
# install.packages("pak")
pak::pak("gmontaletti/skillviz")Main features
| Domain | Key functions |
|---|---|
| Data loading |
read_esco_mapping(), read_isco_groups(), deduplicate_annunci()
|
| Classification crosswalks |
build_cpi_esco_crosswalk(), prepare_annunci_esco(), prepare_annunci_geography()
|
| Skill relevance |
compute_balassa_index(), compute_skill_diffusion(), build_skillist()
|
| Co-occurrence networks |
compute_cooc_matrix(), filter_relative_risk(), cooc_all_professions()
|
| Community detection | detect_skill_communities() |
| Professional distance |
compute_profession_distance(), cluster_professions()
|
| Salary analysis |
extract_salary_data(), compute_salary_by_period(), compute_salary_by_skill()
|
| Source stability |
prepare_source_tsibble(), compute_source_features(), filter_stable_sources()
|
| Temporal analysis |
compute_skill_ranking_series(), compute_skill_variation()
|
| Visualization |
plot_cooc_graph(), plot_skill_ranking()
|
Included datasets
The package ships with three pre-computed datasets:
-
isco_gruppi– ESCO ISCO group classification mapping -
skillist– Master skills list with metadata (frequency, reuse type, domain flags) -
cooccorrenza– Pre-computed global skill co-occurrence network
Quick example
library(skillviz)
library(data.table)
# 1. Load included data -----
data(cooccorrenza)
data(skillist)
# 2. Build co-occurrence graph -----
graph <- plot_cooc_graph(cooccorrenza, top_n = 50)