Purpose

This study aims to develop a building renovation duration prediction model incorporating both scope and non-scope factors.

Design/methodology/approach

The study used a questionnaire to obtain basic information relating to identified project scope factors as well as information relating to the impact of the non-scope factors on the duration of building renovation projects. The study retrieved 121 completed questionnaires from construction firms on tertiary education trust fund (TETFund) building renovation projects. Artificial neural network was then used to develop the model using 90% of the data, while mean absolute percentage error was used to validate the model using the remaining 10% of the data.

Findings

Two artificial neural network models were developed – a multilayer perceptron (MLP) and a radial basis function (RBF) model. The accuracy of the models was 86% and 80%, respectively. The developed models’ predictions were not statistically different from those of actual duration estimates with less than 20% error margin. Also, the study found that MLP models are more accurate than RBF models.

Research limitations/implications

The developed models are only applicable to projects that suit the characteristics and nature of the data used to develop the models. Hence, models can only predict the duration of building renovation projects.

Practical implications

The developed models are expected to serve as a tool for realistic estimation of the duration of building renovation projects and thus, help construction project managers to effectively plan and manage it.

Social implications

The developed models are expected to serve as a tool for realistic estimation of the duration of building renovation projects and thus, help construction project managers to effectively plan and manage it; it also helps clients to effectively benchmark projects duration and contractors to accurately estimate duration at tendering stage.

Originality/value

The study presents models that combine both scope and non-scope factors in predicting the duration of building renovation projects so as to ensure more realistic predictions.

中文翻译：

---

建筑改造项目的建模工期

目的

本研究旨在开发一个包含范围因素和非范围因素的建筑改造持续时间预测模型。

设计/方法论/途径

该研究使用问卷调查来获取与已确定的项目范围因素相关的基本信息以及与非范围因素对建筑改造项目工期的影响相关的信息。该研究从建筑公司中检索了 121 份关于高等教育信托基金 (TETFund) 建筑改造项目的完整问卷。然后使用人工神经网络使用 90% 的数据开发模型，同时使用平均绝对百分比误差使用剩余 10% 的数据验证模型。

发现

开发了两种人工神经网络模型——多层感知器（MLP）和径向基函数（RBF）模型。模型的准确率分别为 86% 和 80%。所开发模型的预测与实际持续时间估计没有统计差异，误差幅度小于 20%。此外，研究发现 MLP 模型比 RBF 模型更准确。

研究局限性/影响

开发的模型仅适用于适合开发模型所用数据的特征和性质的项目。因此，模型只能预测建筑改造项目的持续时间。

实际影响

所开发的模型有望成为实际估计建筑改造项目工期的工具，从而帮助建筑项目经理有效地规划和管理它。

社会影响

所开发的模型有望成为实际估计建筑改造项目工期的工具，从而帮助建筑项目经理有效地规划和管理项目；它还可以帮助客户有效地衡量项目工期，帮助承包商在招标阶段准确估计工期。

原创性/价值

该研究提出了结合范围因素和非范围因素来预测建筑改造项目持续时间的模型，以确保更切合实际的预测。