Hierarchical Machine Learning-Based Growth Prediction Model of Panax ginseng Sprouts in a Hydroponic Environment

Area of Science:

• Agricultural Science
• Biotechnology
• Machine Learning

Background:

• Panax ginseng sprouts are gaining attention for their health benefits, driven by saponin content.
• Optimizing hydroponic cultivation is crucial for increasing ginseng sprout production and reducing cultivation time.
• Existing research lacks focus on early disease prediction and productivity enhancement in hydroponically grown ginseng sprouts.

Purpose of the Study:

• To develop and evaluate a hierarchical machine learning model for predicting the growth outcomes of Panax ginseng sprouts.
• To assess the model's ability to classify germination and rottenness, and predict leaf number and stem length.
• To identify optimal hydroponic conditions for enhanced ginseng sprout cultivation.

Main Methods:

• Ginseng sprouts were cultivated under four distinct hydroponic conditions.
• Physical properties were measured, and environmental data collected using sensors.
• Hierarchical machine learning models (ANN, SVM, Random Forest) were employed for classification and regression tasks.

Main Results:

• The germination classification model achieved an F1-score of approximately 99% weekly.
• The rottenness classification model's accuracy improved from 83.5% to 98.9%.
• Leaf number prediction showed an nRMSE decrease of 33% by week 3, with stem length prediction performing even better.

Conclusions:

• The proposed hierarchical machine learning algorithm effectively predicts germination and rottenness in ginseng sprouts.
• The model demonstrates high accuracy in forecasting key growth parameters like leaf number and stem length.
• This approach offers a valuable tool for early disease detection and productivity improvement in hydroponic ginseng cultivation.