Reviewer has chosen to be Anonymous

Summary of paper in a few sentences:

Overall, the study is an interesting application of marketplace data, but I do not feel that it supports the idea that the constructed CPI offers benefits compared to official CPI measurement and it lacks novelty and significant contribution to the existing literature.

Reasons to accept:

N/A

Reasons to reject:

This is an interesting application of web scraped data for the purpose of building an alternative CPI measurement for Indonesia. I found the paper well written where most of the steps are simply and detailed described. However, although the research has merits it lacks novelty and significant contribution. This is apparent also from the relevant literature part where previous efforts are described and it is not clear what is the significant difference of this study. I am guessing that the position of the study is that the analysis is on the city level.

I understand the importance of the marketplace information as e-commerce volume accounts for a significant part of the overall transactions and it may allow for a more timely and costless analysis. However, I cannot see how the research proves that the constructed CPI is more accurate or at least with the same accuracy as the BPS-statistics CPI. The fact that the constructed index maybe more volatile does not mean that is more accurate. The comparison among the two indices is quite superficial and is based only on plotting the values of different commodities. We cannot see a statistically significant difference in the results and most importantly we cannot understand which index is most accurate. Another shortcoming of the study is the representativeness of the data as one specific online marketplace is used. In addition, from the table 3 we can see that in most cases the BPS Statistics coverage is higher than the marketplace data. Regarding Table 3 it will be beneficial to clarify what you mean with the term final coverage. I am guessing is the number of commodities that are present in both BPS Statistics and Marketplace data. In that case the sample is also significantly reduced.

Overall, the study is an interesting application of market data, but I do not feel that it supports the idea that the constructed CPI offers benefits compared to official CPI measurement and as I have aforementioned I do not find a significant contribution to the existing literature.

Nanopublication comments:

Further comments:

Reviewer has chosen to be Anonymous

Summary of paper in a few sentences:

The paper is interesting and very relevant in today’s business world. Using web scraping techniques, the paper compares the marketplace Consumer Price Index (CPI) and Badan Pusat Statistik (BPS)-Statistics CPI in Indonesia.

Reasons to accept:

The paper relates well with the existing literature. Key concepts are also presented clearly. The methodology is novel and interesting.
In general, the paper is well-written and clearly communicated. But the present and past tenses are mix-used in the manuscript. Please amend for consistency. The paper will benefit from a final round of proofreading to eliminate typographical and grammatical errors (e.g., p.3). The conclusion in the paper needs to be strengthened. A more critical reflection on the research limitations and future work is needed. It can also be very helpful if some practical implications can be provided in the paper.

Reasons to reject:

None

Nanopublication comments:

Further comments:

Overall, the paper is likely to be of some interest to the Data Science audience.

By Cristina-Iulia Bucur https://orcid.org/0000-0002-7114-6459

Review Details

---

Reviewer has chosen not to be Anonymous

Summary of paper in a few sentences:

As digital economy and particularly, e-commerce, seems to experience a rapid growth in recent years in Southeast Asia, in this article the authors propose to include data from the online marketplaces available in Indonesia to compute commodity-level and expenditure group-level CPI (Consumer Price Index). Moreover, the authors propose to compare these two computed CPIs from the online marketplaces to the CPI computed using the data from BPS-Statistics Indonesia, a non-ministerial government agency that conducts various studies, among which is the calculation of CPI per commodity baskets per city. This comparison between different CPI calculations (from the online marketplaces to the statistics collected by onsite retailers by BPS) is interesting in the context of the rapidly evolving online market and also to see if and how the e-commerce affects the CPI.

Reasons to accept:

- the general idea of the article is interesting, especially in the context of the evolving e-commerce market.
- the presentation of the general and some particular workflows (the "...Flow" diagrams as they are titled in the article) is clear, despite having some that do not provide a lot of information, like the ones in Figure 3 and 4.
- the data used is available in a github repository.
- the authors seem to undersell their results and work by presenting and structuring them in a way that seems lacking. If more explanations and interpretations would be given for certain figures, tables and the results presented there, also showing why they are important and what they show, that would improve the article a lot.

Reasons to reject:

- in the end the authors seem to fail to present the results in a clear manner and the discussion and interpretation about these results seems to be lacking. For example, despite presenting/emphasizing the "red chilli pepper" example in Malang in Table 4, it would have been interesting to see why this is the case and also discussing some of the other commodity cases instead of mentioning the value of the BPS-Statistics CPI vs. the Marketplace CPI (43.86 vs. 60.97), numbers which do not mean a lot as long as they are not put into context; what do these numbers mean, what does the difference between them mean? Are they expected in the economical and "cultural" context of Indonesia, for example?
- some revisions in terms of the usage of English language might be needed. Some sentences lack a verb while the use of articles ("a/an" and "the") is completely missing in some sentences.
- the figures and tables presented are sometimes confusing, also the use of the English language needs to be reviewed as sometimes the explanations (when present) are confusing due to the language used. Also, the presence of some figures might not be necessary as they are now and they seem to be confusing or not have any added value as long as they are not put into context or have more explanations attached to them; e.g. Figure 2 and Figure 8.
- in Table 1 the differentiator and the explanations are unclear at times.
- values from Table 3 require more explanations as it is not clear what the numbers mean; if the emphasis is on the comparison between the BPS-statistics and the Marketplace Coverages, then this needs to be explained in text as well. Also, what is the Final Coverage column and what does it represent?
- Figure 9 is confusing. You introduce the notion of "movement pattern" without explaining what this is and why it is important and the explanation of this figure seems to be lacking. Also, how the aggregate numbers of events are computed is confusing. The subfigures need some explanations or translation (what does the oy axis represent?!).The conclusion ("This means that the movement of marketplace data consumption is more volatile than the pattern of consumption values obtained from BPS-Statistics") illustrates very well the purpose of the figure and what it wants to show, but I felt it could have been explained maybe from the start to show why such a figure would be important.
- the legend of Figures 10, 11 and 12 seems a bit "out of focus", meaning, it can be put somewhere more visible. Also, some inconsistencies are present in some figures, for example, why are the numbers from BPS-Statistics in blue, while the ones from the Marketplace CPI are in orange (the color of BPS-Statistics column)?
- some references need to be revised (see, for example, in the "Data Collection" -> "Documents" section; the references mentioned in points 3 and 4 seem to be wrong, maybe references [15] and [18] should be there instead of [14] and [15]).

Nanopublication comments:

Further comments:

The article needs major revisions to be accepted, in my opinion. I think the following need special attention: the use of English language, the figure and table presentations (some of which might need to be removed if they do not bring useful information), adding more explanations and interpretations to the results.