What This App Does
This app predicts consumer liking scores for papaya fruit based on metabolite concentrations, using an XGBoost machine learning model trained on sensory panel data from 125 untrained consumers across nine distinct papaya genotypes.
Important Limitations
Predictions are indicative rather than definitive and may not generalise beyond the genotypes, growing conditions, and analytical methods used in the original study. Metabolite concentrations should be collected using certified reference standards and appropriate internal reference standards following the published methodology — deviations may reduce accuracy. These models do not require matrix correction. Predicted liking scores reflect the average preferences of the training panel and may not represent other consumer groups or demographics. This tool is intended to support research and decision-making, not replace sensory evaluation.
For full methodological details, see the published article: Metabolomic modelling of sensory characteristics and consumer liking in papaya fruit.
Step-by-Step Instructions
- Select a Model: Use the sidebar to choose 'All Papaya' or 'Red-Fleshed Only'.
- Prepare Your Data: Download the template below and fill in your metabolite concentrations.
- Enter Data: Go to 'Data Input' and either type values manually or upload your completed template.
- Validate: Click 'Validate Data' to check for issues.
- Predict: Go to 'Predictions' and click 'Run Prediction'.
- Download: View results and download as Excel.
Download Template
Template with correct column names for the selected model:
Download Template (.xlsx)Acknowledgements
This project was funded by Hort Innovation and the Australian Government, through Hort Frontiers Fund AS19003 and the Griffith University Postgraduate Research Scholarship. All analytical chemistry assays were undertaken with the trusted oversight of Alan White, Griffith University Analytical Facility Manager. All food sensory analysis was managed by Emma Hassal, Queensland Alliance for Agriculture and Food Innovation Nutrition and Food Science Senior Research Assistant and Laboratory Manager.
Training Data Ranges
Metabolite concentration ranges from the training data. Amber rows show where your input values fall outside these ranges.
What is XGBoost?
XGBoost (eXtreme Gradient Boosting) is a machine learning algorithm that combines many simple decision trees, each correcting the mistakes of the previous one. It has learned the relationship between metabolite concentrations and consumer liking scores from papaya sensory data.
Model Performance
Feature Importance
How much each metabolite contributes to the prediction:
Model Last Updated: February 2026
The papaya flavour wheel describes the range of aroma and taste descriptors identified in papaya fruit by trained sensory panellists. The inner ring groups the sensory categories; the middle ring are the traits used to score fruit samples; the outer ring descirbes the middle ring traits
