Food Composition Nutrient Matrix from TKPI 2017 and USDA SR Legacy 2018

Two harmonized food composition datasets from TKPI 2017 and USDA SR Legacy 2018 reformatted as wide-format nutrient matrices. The repository, created by ULM-DS-Lab, contains data reformatted for machine learning applications with a shared set of 18 common nutrients. It was last updated on the platform in March 2026.

Use Cases

• Nutrient prediction based on food composition features
• Food type classification based on nutrient profiles
• Missing value imputation for nutrient data
• Dietary analysis based on harmonized nutrient matrices
• Cross-dataset generalization experiments based on 18 common nutrients

Strengths

• Harmonized set of 18 common nutrients enabling cross-dataset experiments
• Combines data from two authoritative sources: TKPI 2017 and USDA SR Legacy 2018
• Reformatted specifically for machine learning applications

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Description metadata is limited; actual data quality requires manual inspection after download

Provenance

Source

ULM-DS-Lab, combining TKPI 2017 and USDA SR Legacy 2018 data

Collection Method

Reformatted from original sources into wide-format nutrient matrices

Time Range

Data from 2017 and 2018 sources

Freshness

Last updated 2026-03-25 01:39:28; freshness should be verified

Geography

Region 'us' indicated by platform tags