Phenotypic evolution of broomcorn millet (Panicum miliaceum) in ancient Turpan, Xinjiang, China

The analysis scripts and data supporting this study are provided in this repository to ensure full reproducibility of the classification and temporal analyses. The repository is structured as follows: Three Excel files serve as the primary data inputs: Weedy.xlsx and Cultivated.xlsx contain morphological measurements of modern weedy and cultivated broomcorn millet, used for training the discriminant model; Ancient.xlsx contains measurements of archaeological millet remains from Turpan, used for classification and outlier detection. Four Python scripts implement the core analytical workflow: weedy_vs_cultivated_analysis.py performs the binary discrimination analysis between weedy and cultivated millet, generating feature-separation figures, classification metrics, and model comparison outputs. classify_ancient.py loads the training data and Ancient.xlsx, applies the trained classifier to ancient samples, and performs outlier detection using isolation forest. combine_predicted_csv_analysis.py reads the predicted CSV outputs from the classification step and generates the combined ancient analysis report. turpan_analysis.py conducts the diachronic temporal analysis of the Turpan millet measurements, including summary statistics and regression trends. All analyses were implemented in Python 3.11.11, with key dependencies including NumPy 1.26.4, pandas 2.2.2, matplotlib 3.9.0, seaborn 0.13.2, scikit-learn 1.8.0, and openpyxl 3.1.5. To reproduce the results, install the required dependencies in a Python virtual environment and execute the four scripts in sequence from the repository root. Outputs, including figures and summary tables, are written to the corresponding output directories (analysis_output*, Ancient_Classification_Results/, turpan_analysis_output/). The code uses relative paths from the project directory and includes early checks for missing input files.