Meta-IP: An Imbalanced Processing Model Based on Meta-Learning for IT Project Extension Forecasts

Abstract

With increasing developments in information technology, IT projects have received widespread attention. However, the success rate of large information technology projects is extremely low. Most current extension forecast models are designed based on a balanced number of samples and require a large amount of training data to achieve an acceptable prediction result. Constructing an effective extension forecast model with a small number of actual training samples and imbalanced data remains a challenge. This paper proposes a Meta-IP model based on transferable knowledge bases with few-shot learning and a model-agnostic meta-learning improvement algorithm to solve the problems of sample scarcity and data imbalance. The experimental results show that Meta-IP not only outperforms many current imbalance processing strategies but also resolves the problem of having too few samples. This provides a new direction for IT project extension forecasts.