Scholarlink

<p>This paper proposes a strawberry yield prediction scheme suitable for non-parametric regressions, using data collected by developed Internet of Things (IoT) nodes built from general-purpose devices. That is, we propose a method to predict the yield of strawberries, as one example of fruit crops harvested multiple times in one season. Specifically, we devise a model in which crop yields are affected by short-term environmental fluctuations and propose introducing differential values in regression analysis. Using four years of IoT and harvest data on strawberries grown in greenhouses by local farmers, we evaluate the yield prediction accuracy of the proposed method using the mean absolute percentage error (MAPE) and the adjusted coefficient of determination (<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="43926_2025_219_Article_IEq1.gif" Format="GIF" Height="23" Rendition="HTML" Resolution="72" Type="Linedraw" Width="30" /> </InlineMediaObject> <EquationSource Format="TEX">\(R_{\textrm{adj}}^{2}\)</EquationSource> </InlineEquation>). As a result, the proposed scheme reduces MAPE by up to 50% (0.36 to 0.18) and improves <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="43926_2025_219_Article_IEq1.gif" Format="GIF" Height="23" Rendition="HTML" Resolution="72" Type="Linedraw" Width="30" /> </InlineMediaObject> <EquationSource Format="TEX">\(R_{\textrm{adj}}^{2}\)</EquationSource> </InlineEquation> by 20% (0.60 to 0.72) compared with the conventional method, demonstrating substantial gains in prediction accuracy. In particular, the yield prediction accuracy is the highest when using generalized additive model (GAM) regression, a type of non-parametric regression. Additionally, we show that the proposed method has a high affinity for GAM regression.</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

A strawberry yield prediction scheme suitable for non-parametric regression using general-purpose IoT nodes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月17日

查看全文
  • Satoshi Yamazaki,
  • Masaki Aono,
  • Yoshikazu Kiriiwa

摘要

This paper proposes a strawberry yield prediction scheme suitable for non-parametric regressions, using data collected by developed Internet of Things (IoT) nodes built from general-purpose devices. That is, we propose a method to predict the yield of strawberries, as one example of fruit crops harvested multiple times in one season. Specifically, we devise a model in which crop yields are affected by short-term environmental fluctuations and propose introducing differential values in regression analysis. Using four years of IoT and harvest data on strawberries grown in greenhouses by local farmers, we evaluate the yield prediction accuracy of the proposed method using the mean absolute percentage error (MAPE) and the adjusted coefficient of determination ( \(R_{\textrm{adj}}^{2}\) ). As a result, the proposed scheme reduces MAPE by up to 50% (0.36 to 0.18) and improves \(R_{\textrm{adj}}^{2}\) by 20% (0.60 to 0.72) compared with the conventional method, demonstrating substantial gains in prediction accuracy. In particular, the yield prediction accuracy is the highest when using generalized additive model (GAM) regression, a type of non-parametric regression. Additionally, we show that the proposed method has a high affinity for GAM regression.