A new attribute-linked residential property price dataset for England and Wales, 2011–2019

This paper describes a sound and logical process that creates a valuable housing dataset. By linking LR-PPD and EPC data, countrywide housing transaction records being enriched with highly useful floor area attributes. Such a dataset is very much welcomed with a high matching rate and open access code. The benefits of this process are evident with stats figures provided in the summary.

I found the paper is well organized and presented. It is easy to follow although a very complex process was described. The diagrams illustrate the logic behind the steps well. This makes the matching results justifiable.

I have the following questions or suggestions which hopefully will help to improve it further:

1. This paper, as it is titled, is a data description summary. It will be great if it can be extended into a method paper where more details about the rules can be included.
2. Relate to this, I find the paper describes the logic of data processing very well. But there are limited examples provided.  For instance, in P5, 95 new variables in EPC and 180 variables in LR-PPD were mentioned to be included. Would be better to see a couple of examples. Also, I appreciate there are 251 matching rules they are detailed and complex. It would be great to see some examples too. This will make the logic clearer. At the moment the paper is rather conceptual.
3. In terms of validations, it would be great if a manual random check of the matching results can be included. This will introduce the data with extra (a) examples of matching results (b) accuracy descriptions at the end.
4. After aggregating to the census unit, whether it is possible to compare with the census housing figures such as the distribution of house types or other types of commonly sorted attributes?
5. As mentioned, PPD data is updated regularly. It may be worth checking different versions of data to see if they result in different matching rates. This will give us a reliability test.
6. It may be also worthwhile to include more limitations discussed at the end.

Thanks to Bin and others for sharing this interesting article and generating a useful dataset. Hope more details of the method and wider application will be made available soon.

The lack of house price per square metre data has caused serious problem in comparing the UK and international housing markets and proposing suitable policies to tackle the domestic housing crisis.

The publication presents an exciting academic research in this area and filled an important research gap well documented by reviewing related literature on developing and applying related house price datasets.

Based on technically advanced linking and cleansing methods, the research illustrats how to generate a comprehensive house price dataset covering around 90% properties including flats within a house. It also takes account of various property attributes like energy efficiency. Meanwhile, it illustrates in great details how to link, clean and update data which paves a way for wider academic use.

One highlight of this publication is its state-of-art data linkage method. It involves a matching method containing a four-stage (251 matching rules) linking various sources with algorithm testing the matching efficiency.

The publication is also well-written with properly displayed figures and logically organised chapters. The authors manage to illustrate the complicated process via a workflow figure which greatly facilitates understanding of an interested academic.

While it is quite difficult to pick a weak point from the publication, some suggestions are provided if the authors are interested in further related research. Instead of manual correction, there might be possibility of adopting some natural language processing or other text analytical tools to deal with name mismatch issue mentioned in Section 5. The potential automation process may not only improve efficiency but also provide opportunities for text-based attribute analysis. Secondly,the authors can point a lot of areas in which this research can be used. For example, a detailed up-to-date analysis of energy efficiency analysis can be based on the updating data set. The data sets also generate great potential for international comparation studies.

In summary, the publication clearly presents an innovative house price dataset and technically advanced methodology, which will generate huge research impact on the related research area.