Scholarlink

<p>Since the advent of data analytics, researchers found hybrid models combining statistical and artificial intelligence tools to be quite useful in predicting future outcomes, particularly, for binary situations. Classification tree (CT) combined with artificial neural networks (ANN) named as CT-ANN model, is one such popular models which is proven to be quite robust. However, if the parameters are correlated then these models, as such, are not very useful. For repeated measures data, where the chances of correlation being present between parameters is quite high, no useful technique has been proposed so far. We try to modify here the CT technique to filter out correlated features and then use ANN to finally predict binary outcomes. A theoretical proof of the proposed modified CT algorithm is provided. Further we applied the modified CT-ANN model to simulated data and subsequently to real life data and showed the efficacy of the proposed model compared to the CT-ANN model for correlated features data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modified CT-ANN model for correlated repeated measurement data

  • 期刊论文

  • 发表日期：

    2026年4月18日

查看全文
  • Ashis Kumar Chakraborty,
  • Saikat Adhikary

摘要

Since the advent of data analytics, researchers found hybrid models combining statistical and artificial intelligence tools to be quite useful in predicting future outcomes, particularly, for binary situations. Classification tree (CT) combined with artificial neural networks (ANN) named as CT-ANN model, is one such popular models which is proven to be quite robust. However, if the parameters are correlated then these models, as such, are not very useful. For repeated measures data, where the chances of correlation being present between parameters is quite high, no useful technique has been proposed so far. We try to modify here the CT technique to filter out correlated features and then use ANN to finally predict binary outcomes. A theoretical proof of the proposed modified CT algorithm is provided. Further we applied the modified CT-ANN model to simulated data and subsequently to real life data and showed the efficacy of the proposed model compared to the CT-ANN model for correlated features data.