# Leveraging TIS-Enhanced Crayfish Optimization Algorithm for High-Precision Prediction of Long-Term Achievement in Mathematical Elite Talents

## TL;DR

This paper introduces a new algorithm inspired by crayfish behavior to predict long-term success in mathematically talented students.

## Contribution

The novel TIS-enhanced Crayfish Optimization Algorithm improves hyperparameter optimization for predicting elite mathematical achievement.

## Key findings

- The TISCOA algorithm outperforms other metaheuristic optimizers in predicting long-term mathematical achievement.
- Non-cognitive factors like Emotion Regulation significantly influence long-term outcomes.
- Temporal variables such as the Latency Period shape developmental trajectories in elite talents.

## Abstract

Traditional talent identification systems often rely on static assessments and overlook the dynamic nature of long-term development. To address this limitation, this study proposes a biomimetic predictive framework inspired by crayfish behavioral ecology. The Crayfish Optimization Algorithm (COA), derived from adaptive foraging and competition mechanisms observed in crayfish, is enhanced through a Thinking Innovation Strategy (TIS) to form TISCOA for hyperparameter optimization of a Gradient Boosting Decision Tree model. Using a five-year longitudinal dataset of 160 elite mathematical students, the framework models Professional Achievement in Mathematics (PAM) from multidimensional baseline indicators. Comparative experiments with multiple metaheuristic optimizers show that the proposed approach achieves stable generalization performance within the examined cohort. Feature attribution analysis indicates that non-cognitive factors, particularly Emotion Regulation, contribute substantially to long-term outcomes, while temporal variables such as the Latency Period further shape developmental trajectories. Residual analysis highlights heterogeneous patterns that may reflect unobserved contextual influences. Overall, the study demonstrates how a biologically inspired optimization mechanism can support interpretable and stability-oriented longitudinal prediction in small-sample educational settings.