**5. Conclusions**

This study proposes and applies a systematic data analytic methodology to analyse the experimental data obtained from tests of HPC samples with different admixtures. In contrast with other relevant studies aimed primarily at performing experiments (and providing rationale for the experiments based on materials, physics, or chemistry) or identifying the optimal HPC admixture(s) for grouting concrete materials to be used in the sea (e.g., for the base construction of offshore wind turbines), the purpose of this study is to perform a thorough investigation of the experimental variables related to the testing data.

To achieve this purpose, a methodological framework is proposed. In order to generate comprehensive and in-depth views of the data, numerous methods are utilised, including *P-Co-Co, Cos-Sim*, SLR, and heatmap or heat-based tabularised visualisation. This approach is significantly different from those of other experimental-based research or admixtureselection studies. Highlights of the research activities, results, and insights gained are presented below.


transformation was confirmed by another eight correlation matrices with variable *X* in Equation (2) being converted using transforms (2)–(9) described in Section 2.3.4.


The methods used in this study are common, and utilising them for this study was proven to be effective. Therefore, it should also be valid to utilise the methodological framework proposed in this research for similar purposes, e.g., data analysis for other concrete samples with different admixtures. From this perspective, a future story upon taking the proposed framework is perhaps clear from scratch. When a construction project is launched, one first sees if the HPC material(s) is to be used.

As can be expected, if so (using HPC) and if the planned admixtures of the HPC samples to be tested are analogous to the ones being tested in this study, efforts related to testing the materials can be reduced considerably by either policy. The first policy is to use the direct testing results for the samples in the established knowledge base (if 'no testing' is allowed legally). The second is to determine the testing item to be 'predicted' (as desired), look up the knowledge base to know what any other testing item can predict it (and also the information about whether the prediction will be effective and/or accurate), and decide whether to really but safely save the money, and time in most of the cases, for making tests by anticipating the results for the testing item (if some regulated testing items are mandatory).

However, if so (using HPC) and the planned HPC samples to be tested are different (e.g., due to a diversified purpose of use), or if not so (using the non-HPC material), theoretically, the first step is to make similar tests for these samples, followed by using the proposed framework to conduct the thorough data experiments (in order to identify the relationship and the mutual predictive power between each pair of sample parameters). Then as can be imagined, another knowledge base for the samples included by this purpose of use can be established. Although the outcomes and the unanticipated insights gained may vary in different cases or contexts, after this one-time job, what follows is a similar policy-making problem (see the former paragraph). Anyhow, all this may support efforts to save the time required for pre-project experiments, especially when the project is urgent, which is a crucial but inevitable fact in the current A/E/C industry.

It should be noted that the applications of this framework are not limited to cases in which the source datasets for each measure are tested using conventional methods (i.e., a parameter of HPC samples may have more than one testing method). For example, the framework could be applied when a part of the data is obtained using NMR (nuclear magnetic resonance) [48,49]. In many countries, NMR equipment is legal and commercialised, and the data for some variables can be obtained [50,51]. Other than this, except for HPC, the question of 'can similar data analysis using this framework be carried out for other construction materials?' is also worth of exploration.

Future research directions are shown in Figure 5, wherein the aim of this study which is clearly differentiated from other relevant studies.

**Figure 5.** The role of this research for HPC studies.

In the figure, the experimental-oriented studies are focused on the performance of the HPC, the inclusion of parameters for developing experiments, and statistics and variability in the data. A critical element that future research directions rely upon is the testing data sets produced by this research. Following this step, there are clear boundaries among the three research directions. The materials, physics and chemistry studies research direction involves a significant amount of opportunity for future studies. However, the knowledge base constructed for the HPC testing parameters as an outcome of the data analytics studies may benefit advance data studies in the future as well as subsequent selection decision modelling studies and current cementitious material portfolio determination studies. The selection decision modelling studies aim to rank the HPC samples with different admixtures and select the optimal candidate using the scientific decision models; therefore, the knowledge base may provide a precise numerical foundation for these models (and added value from the expert knowledge of the researchers).

**Author Contributions:** Conceptualization, methodology, software, visualization, Z.-Y.Z.; data curation, investigation, supervision, validation, W.-T.K.; funding acquisition, project administration, writing—original draft preparation, writing—review and editing, Z.-Y.Z. and W.-T.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Ministry of Science and Technology (MOST, Taiwan, ROC), grant number MOST 110-2410-H-992-020 and 110-2637-E-992-001. The APC was funded by MDPI discount voucher and MOST 111-2410-H-992-011. Note that the institution 'Ministry of Science and Technology (MOST)' has been renamed as National Science and Technology Council (NSTC) in August 2022 in this area. However, we still keep the original funding number for these projects were approved before the transition.

**Data Availability Statement:** N/A. All related data for the results of this study are detailed in the appendix sections.

**Conflicts of Interest:** The authors declare no conflict of interest.
