Developing, evaluating and reviewing ML-based predictive model for numerical databases in civil engineering

Over the past decade, the application of machine learning techniques in both industrial and academic projects has become common practice. During this period, hundreds of machine learning-based predictive models have been developed for various topics using various machine learning techniques. Unfortunately, some of these models are inefficient, inaccurate, or unusable. This is due to a lack of data science knowledge among some authors or even reviewers. This paper aims to provide guidance for authors on model development and evaluation, and a checklist for reviewers on machine learning-based predictive models. The paper discusses in detail the basic steps for developing, evaluating, and reviewing machine learning-based predictive models. These steps include: data collection (size, type, and source); data preprocessing, forming a valid database, and partitioning; statistical analysis, correlation, and sensitivity; technique selection, model training, and performance evaluation; and model comparison, discussion of results, and drawing conclusions. In addition, the paper provides an overview of available research quality checklists and proposes a more detailed and general checklist for evaluating and reviewing predictive models.