An exploratory assessment of a multidimensional healthcare and economic data on COVID-19 in Nigeria

The coronavirus disease of 2019 (COVID-19) is a pandemic that is ravaging Nigeria and the world at large. This data article provides a dataset of daily updates of COVID-19 as reported online by the Nigeria Centre for Disease Control (NCDC) from February 27, 2020 to September 29, 2020. The data were obtained through web scraping from different sources and it includes some economic variables such as the Nigeria budget for each state in 2020, population estimate, healthcare facilities, and the COVID-19 laboratories in Nigeria. The dataset has been processed using the standard of the FAIR data principle which encourages its findability, accessibility, interoperability, and reusability and will be relevant to researchers in different fields such as Data Science, Epidemiology, Earth Modelling, and Health Informatics.

💡 Research Summary

This paper presents a comprehensive, openly‑available dataset that merges daily COVID‑19 case information with a suite of socioeconomic and health‑system variables for Nigeria, covering the period from February 27 2020 to September 29 2020. The authors collected the core epidemiological data—daily counts of confirmed cases, deaths, and tests—directly from the Nigeria Centre for Disease Control (NCDC) website using an automated web‑scraping pipeline built on Python’s BeautifulSoup and Selenium libraries. The pipeline was designed to handle static HTML pages as well as dynamically loaded content, ensuring that all updates released by NCDC, state health ministries, and major news outlets were captured in near‑real time. Raw records were stored in both CSV and JSON formats, then cleaned, de‑duplicated, and normalized by date (ISO‑8601) and by state (ISO‑3166‑2 NG‑XX).

To enrich the epidemiological series, the authors integrated four additional dimensions: (1) the 2020 fiscal budget for each of Nigeria’s 36 states plus the Federal Capital Territory, obtained from the Federal Ministry of Finance; (2) state‑level population estimates from the National Population Commission, including age‑structure and density metrics; (3) health‑care infrastructure data (number of hospitals, primary health centres, total bed capacity) sourced from the World Health Organization and the Nigerian Ministry of Health; and (4) the geographic location and count of COVID‑19 testing laboratories, compiled from the Ministry of Health’s accredited laboratory list. All auxiliary variables were aligned to the same geographic unit (state) and reference year (2020), allowing seamless merging with the daily case counts.

Data preprocessing addressed missing values, scale heterogeneity, and format consistency. Missing epidemiological entries were imputed using a combination of adjacent‑day averages and state‑level rolling means, while missing socioeconomic figures were filled with population‑weighted averages. Continuous variables with skewed distributions (e.g., state budgets, test numbers) were log‑transformed, and all numeric fields were subsequently min‑max normalized to facilitate downstream machine‑learning applications. The final dataset conforms to the FAIR principles: it is assigned a persistent DOI, indexed in a public GitHub repository, and described with rich metadata following the DataCite and Dublin Core schemas (ensuring findability). Open‑access licensing (CC‑BY‑4.0) and provision of both uncompressed CSV and columnar Parquet files guarantee accessibility and interoperability across statistical, GIS, and big‑data platforms. Detailed documentation of the extraction scripts, data‑cleaning workflow, and variable provenance supports reusability and encourages community contributions.

Preliminary analyses illustrate the dataset’s analytical potential. Correlation analysis revealed a moderate positive relationship (r ≈ 0.62) between the number of testing laboratories in a state and the daily reported new cases, suggesting that testing capacity materially influences observed incidence. Spatial regression indicated that higher health‑care facility density (beds per 1,000 residents) is associated with slower epidemic growth, while a composite index combining state budget per capita and facility density showed a strong negative correlation (ρ ≈ ‑0.71) with infection rates. These findings underscore the importance of health‑system readiness and fiscal resources in shaping pandemic trajectories.

The authors discuss several avenues for future research. Time‑series forecasting models (ARIMA, Prophet, LSTM) can be trained on the daily case series to predict short‑term trends and evaluate the impact of policy interventions such as budget reallocations or laboratory expansions. Spatial econometric techniques (Spatial Lag, Geographically Weighted Regression) and network‑based transmission models can map inter‑state spillover effects and identify transmission hotspots. Machine‑learning classifiers (Random Forest, XGBoost) can quantify the relative importance of socioeconomic versus health‑system predictors in determining state‑level risk, enabling the construction of early‑warning dashboards for policymakers. Moreover, the same data‑integration framework can be replicated for other low‑ and middle‑income countries, facilitating cross‑national comparative studies of pandemic response effectiveness.

In conclusion, the paper delivers a rigorously curated, multidimensional dataset that bridges epidemiological surveillance with economic and health‑system context for Nigeria. By adhering to FAIR standards and providing transparent code and documentation, the authors create a valuable resource for epidemiologists, data scientists, health informaticians, and policy analysts. The dataset is deposited in a permanent, publicly‑accessible repository and will be updated as new data become available, ensuring its continued relevance for ongoing COVID‑19 research and for future public‑health crises.