Data analysis of the current housing situation

[Deo14]

This is a good first attempt at data gathering.

The initial analysis is a bit cumbersome. For example, the words "considerably worse" are judgement calls, not data descriptors.

The use of the word "demand" is a two-edged sword. "Supply and demand" works well enough on an economic scale, but, when combined with "considerably worse", may be interpreted as a value judgement call.

"There are more FC houses in Japanese servers than in North American servers" is a better factual statement based on the data you've derived.

"Undesirable plots" is, again, a value judgement, not a factual statement.

Several of the statements are opinion which cannot be inferred from the data. "It's healthier" is meaningless in the context of the data. The use of the term "servers" should be replaced by "Worlds", as no World consists of a single physical server, and each housing Ward consists of more than one physical server.

"Most popular" and "prestigious" are also value judgements not reflected from specific data. It's useful to learn the biases of the analyst.

This was useful: "In entire game, there is 729 000 plots, but only 70 888 free plots. Out of these free plots, over a half (38 020) of them are on the new servers"

This was not: "New servers and wards carry every stat". It is not clear what the statement was supposed to mean.

This statement should never appear in data analysis: "If we ignore new servers ..." as (1) there are new servers and (2) there has to be sufficient reasoning (and making the numbers look pretty is not sufficient) behind ignoring data.

Behaviour is also not useful for data analysis, unless it is one of the defined variables being studied. Better to use "Differences in EU/NA and JP's player numbers provide some hints at why the availability of housing differs by type in each region".

And again, Dynamis has been available for players for 8 months. If the 'player population' site used does not have any relevant data, then the data is useless for the analysis, and should not have been included at all. If there is relevant data, then it should not be excluded simply because it makes the interpretation of results "wonky" (to use an extremely non-technical term).

Either of the solutions proposed would resolve issues. I anticipate that SE's response would be to create more wards.

Thanks for feedback.

"Undesirable plots" is, again, a value judgement, not a factual statement.
.
"Most popular" and "prestigious" are also value judgements not reflected from specific data. It's useful to learn the biases of the analyst.

What is exactly wrong with calling something "most popular", "prestigious" or "undersirable plots"? If we rate demand, isn't that essentially how much is something popular or (un)desirable for people based on data (at least in this case)? Would simply calling it "most demanded" and "least demanded" be more correct?


As for "If we ignore new servers, EU would have nearly 3x less free plots than JP, and NA 2.5x less than JP." - I wanted to give perspective on how did EU/NA get affected by new servers (also this specific point was actually taken from my first version, where it was more clear why I put that in). I should definitely add same thing, but with all servers included, so you can actually see the changes. But thinking back, evaluating this would be better with some timeline with older datasets to really get something meaningful out of that.

And again, Dynamis has been available for players for 8 months. If the 'player population' site used does not have any relevant data, then the data is useless for the analysis, and should not have been included at all. If there is relevant data, then it should not be excluded simply because it makes the interpretation of results "wonky" (to use an extremely non-technical term).

I was taking population data from FFXIVCensus, from # of endgame characters. The fact that Dynamis doesn't have data for it didn't seem that crucial for me, since I've also had analysis per regions. Since Dynamis has enough plots, it seemed fine to just ignore it. Obviously that is not a best solution by any means, but I didn't know how else should I evaluate housing availability without population data.

Anyways, thanks again for feedback. Looking at the data, do you think there is better way to evaluate housing availability, then just simple average of free plots per population?