Synthetic Aviation Spare Parts Demand Inventory and Stockout Dataset for Supply Chain Analytics and Digital Twin Research
Description
This dataset provides a reproducible synthetic panel of aviation spare parts demand, inventory, stockout, flight cycle, and supplier lead time observations for research in aviation maintenance supply chains, spare parts forecasting, inventory analytics, and digital twin applications. The dataset contains 12,960 part location month observations covering 60 synthetic spare parts, three stocking locations, and a 72 month period from January 2018 to December 2023. The parts are divided into fast moving, slow moving, and intermittent demand classes. Monthly demand is generated using a zero inflated Poisson process with a synthetic flight cycle driver incorporating seasonal variation and a mild temporal trend. Inventory states and stockout indicators are generated using a reorder point replenishment simulation, while supplier lead times are sampled from 15, 30, 45, and 60 days. The dataset is accompanied by the exact Python generation script and validation scripts required to reproduce and verify the deposited data. Keywords Use something like: Aviation spare parts Aircraft maintenance Spare parts forecasting Inventory management Demand forecasting Intermittent demand Stockout prediction Supply chain analytics Digital twin Synthetic dataset MRO Aviation maintenance One critical point When Mendeley asks whether the data are synthetic/simulated, select the appropriate option indicating that they are synthetic/simulated. Do not describe this as real airline operational data. Your README and generator clearly establish that it is a synthetic dataset. Once Mendeley gives you the DOI, send me the DOI or the Mendeley metadata page. I can then check the published record against the code/CSV and help you make the final manuscript Data Availability Statement, Code Availability Statement, and repository citation exactly consistent with the DOI.
Files
Steps to reproduce
Download all files associated with this dataset. Install Python 3.10 or later and the required packages: NumPy, pandas, and SciPy. Run the deterministic dataset generator from the directory containing the script: python generate_obj1_synthetic_demand.py --seed 42 --out obj1_synthetic_demand.csv The generator creates a synthetic panel containing 60 spare parts, three stocking locations, and 72 monthly observations from January 2018 through December 2023, producing 12,960 part location month records and 13 variables. The generator assigns 20 parts to each of three demand classes: fast moving, slow moving, and intermittent. Monthly demand is generated using a zero inflated Poisson process scaled by a synthetic flight cycle driver. Inventory states, issued quantities, stockout indicators, and supplier lead times are then generated using the documented simulation procedures in the generator. Validate the generated dataset using: python validate_dataset.py obj1_synthetic_demand.csv The validation script checks record count, missing values, duplicate part location month keys, demand and issue consistency, stockout indicator consistency, number of parts, locations, months, and class allocation. The literature validation utility can be run using: python literature_validation.py obj1_synthetic_demand.csv The canonical deposited CSV was generated using random seed 42. The deposited generator and canonical CSV correspond exactly, allowing deterministic regeneration of the deposited dataset.