Abstract
There is relatively little public and model-ready data on industrial machinery for African economies, making it hard to perform quantitative analysis or to train language models on numeric tasks grounded in that setting. We release two resources to address part of this problem.
The first is the Nigeria Machinery Usage and Failures Dataset, which includes 89 machine-level records across 28 indicators covering Nigeria's manufacturing and oil and gas sectors from 2006 to 2025. Each record cites a public source and is decoded by a codebook.
The second is a method for building chain-of-thought (CoT) reasoning examples from these sparse numeric values, resulting in 94 prompt, completion, and reasoning-trace rows. Each row specifies the real indicator, subsector, year, and source of the record.
The data adaptation work was conducted by Adaption Labs. Along the way, we describe a common problem when using language models to build datasets: prompts may match real numbers without conveying domain knowledge. We demonstrate that fixing this issue increases the share of domain-grounded prompts from 1 out of 78 in an earlier release to 94 out of 94, with every retrieval answer now matching its source value (84 out of 84).
We release the data, the reasoning layer, and a per-row provenance file under CC-BY-4.0. We clarify the limitations: with 89 records and 17 indicators having only one observation, this is a reference and seed dataset, not a large training set, and most reasoning rows are retrieval rather than multi-step computation.
Blogger's Review: The release of this dataset significantly enriches the data resources for industrial machinery in African contexts, especially in terms of potential applications in language model training and quantitative analysis. Despite the limited sample size, the construction of domain-grounded reasoning enhances the data's validity and usability, making it worth attention.