Author: medeaurbanintel

Categories
Japan Workbench

Workbench: Open Datasets

Kendrick Leong

In Workbench we take a deeper look at data transformation for our stories.

In our last post we examined roadside stations, or michi no eki, in Japan, and zoomed in on Kinosato Ten-ei roadside station, in Ten-ei Village, Fukushima Prefecture. To write these stories, we often rely on open datasets, which are downloadable for use, reuse, and modification.  Some examples, specifically from the Kinosato Ten-ei story, include the locations of michi no eki (Japan MLIT, Japanese only) and population censuses (Ten-ei Village, Japanese only). An alternative to open datasets is of course primary data collection, although recent conditions have made going out into the field difficult.

So from the comfort of our homes, we can look to data directories, such as Data.go.jp (English available), which inventories open datasets by Government Agencies, Local Governments, and thematic groups (manufacturing, housing, etc.). Most countries have some kind of open data directory; this one is run by Japan’s Ministry of Internal Affairs and Communications.

Let’s cycle back through the directory’s list of local government databases to find Ten-ei Village’s. Here we find some datasets on population, schools, and disaster resources, such as the locations of AEDs (defibrillators) and evacuation centers. Naturally, data reporting standards and activities vary from organization to organization and government to government, but luckily, we can find datasets in the first place for a village of this size.

For many open datasets, downloaded data takes the form of .csv (comma separated values). These files are read by your spreadsheet software (Excel, Pages, etc.) or data analysis platform (R, Python, etc.). Opening the resident register file for November 2020 (Reiwa 2), we get the following table:

Village SectionHouseholdsMalesFemalesIndividuals
Makimoto5047967941,590
Osato2905005171,017
Hiroto9021,1871,1672,354
Yumoto290280283563
Total1,9862,7632,7615,524
Population by village section, Ten-ei Village, Fukushima Prefecture, November 2020.

Working with Japan open datasets, often the first line of work is translating from Japanese (as was the case with this table).  We can see in the table above that Ten-ei Village is further divided into sections: Makimoto, Osato, Hiroto, and Yumoto. However, this data is but one slice of a larger picture; with population data we are often interested in longitudinal trends – changes over time, and their magnitudes. We took a larger, country-wide look at these trends in our first post.

So let’s bring in not only the data from November 2020, but also for the following months until March 2021 (the last month for which data is available). We are interested in the number of individuals by village section, which will involve some data merging and pivoting. After we attach date fields to each respective dataset, we pivot to a “long” format and get the following table:

Village SectionDateIndividuals
MakimotoNovember 20201,592
MakimotoDecember 20201,590
MakimotoJanuary 20211,583
MakimotoFebruary 20211,579
MakimotoMarch 20211,578
OsatoNovember 20201,024
………
YumotoMarch 2021553
Population by village section, “long” format

Now that we have longitudinal data, spanning the length of November 2020 to March 2021 for each of the Ten-ei Village sections, we can throw it into a graphics software to get a population graph. At MEDEA, we use R and R Studio due to its flexibility and customization options. With just the data we have prepared so far, we get the following graph:

Rather clinical; maybe the type of graph you would see in a research paper.

Next we apply the MEDEA UrbanIntel visual language, including color palettes, fonts, and graph formats. We also add labels for each of the points, but this can be omitted to simplify the graph:

Source: Ten-ei Village Open Data

There is not that much to remark on here; populations stayed relatively constant across the village sections. We can see which sections had more people and which had less, and a miniscule downward trend over the months, with all sections losing residents by March 2021. Alternatively, this data can (and probably should) be presented as a table, which is still able to capture the longitudinal trend:

Village SectionNov ’20Dec ’20Jan ’21Feb ’21Mar ’21
Hiroto2,3552,3542,3552,3512,344
Makimoto1,5921,5901,5831,5791,578
Osato1,0241,0171,0201,0171,013
Yumoto559563557557553
Population by village section, November 2020 to March 2021

So as far as exploratory data analysis with open datasets go, there are many paths from here. We have looked at Ten-ei Village open population data to try to observe longitudinal trends; next we may want to put the Ten-ei Village population in a graph with other village’s populations in Fukushima Prefecture. We may also want to continue looking for open datasets that capture a longer time horizon of the Ten-ei Village population. It would be interesting to see population dynamics for the village over ten, twenty, or even fifty years.

To sum up, open datasets provide a great foundation for exploratory data analysis, as we have walked through in this post. Of course, the exploratory part is limited by what data is available, and it might be a good idea to combine multiple open datasets (Ten-ei Village’s and another village’s, for example). We can also bring in other datasets (migration comes to mind) to connect various piecemeal efforts, made by local governments and government agencies, into a larger data mesh that tells more detailed urban stories.

Categories
Japan Regional revitalization

An Ode to Rest Stops

Kendrick Leong

It’s the crunch of asphalt under your feet after a long drive. It’s the cold mountain air after three hours of recirculated car heating. It’s the coffee milk enjoyed on a picnic bench. Beyond providing the “rest” in rest stops, Japan’s numerous michi no eki, (lit. “roadside stations”), serve a peculiar niche fusing comfort station, farmer’s market, and civic space.

The concept of a highway rest stop is rather simple: somewhere along an expressway to park, refuel (both our stomachs and the gas tank!) and recharge before the next leg of a car journey. But the michi no eki take this concept one step further by showcasing local goods and offering tourism information about the area. These pivots to regional revitalization, with michi no eki as important conduits of community empowerment, have attracted attention from the Japan central government as one strategy to attenuate the effects of dwindling rural populations.

There has been a heartwarming amount written about michi no eki in English, from their catalytic impact on regional revitalization to the individual ingenuity of their operators and the differences between michi no eki and other types of rest areas. Several trends are apparent:

  1. Michi no eki form the skeleton of a decentralized tourism network, giving local entrepreneurs, farmers, and boosters a platform to showcase regional specialties.
  2. Michi no eki are essentially in an arms race to attract travelers, with many boasting hot springs (onsen) on-site and a growing provision of lodging, cultural experiences, and unique attractions.
  3. However, the extent that regional revitalization can pivot off of the flexibility of michi no eki facilities (in offering unique attractions, etc.) may be predicated on, essentially, rural tourism – drawing in guests through regional specialties and experiences (a “can’t get anywhere else” factor).

To the last point, rural tourism is great for promoting regional revitalization (in terms of money injected into the local community and all that). There are also emerging trends that point to regional revitalization beyond mere monetary transactions – that there are sustainable ways to attract interest in rural areas in addition to selling local produce or value-added goods.

One point of pivot could be the michi no eki as civic space (written about extensively in the World Bank’s in-depth exploration of michi no eki), wherein the michi no eki serves as a focal point for knowledge creation (and spillover!) and public services. There are numerous instances where a michi no eki will also serve as a training center, education facility, meeting space, or convention hall, open not only to highway users but also accessible to residents via local roads.

Zoom in: Ten-ei

For many, their first experience in a rural area may be a michi no eki. It was for me – our pit stop on the way to a college club event in Ten-ei village was the area’s michi no eki, Kinosato Ten-ei. Ten-ei village (official site, Japanese), a village true to its name, is a picturesque hamlet in inland Fukushima Prefecture, population skirting 5,000, formed from the merging of several other villages in the mid-20th century. Like many other places around the country, Ten-ei’s population has steadily declined as youngsters leave for larger cities for school or work.

Kinosato Ten-ei, November 2016, Author.

Lacking any passenger railway connections, the main access to Ten-ei is by car, necessitating a rest stop like Kinosato Ten-ei. If there is truly an amenities arms race amongst michi no eki (see: theme park-scale michi no eki such as Kawaba Denen Plaza in Gunma Prefecture) then frankly Kinosato Ten-ei cannot compete. However, it is these smaller michi no eki that have the potential to connect local farmers and entrepreneurs to the vitality of the national expressway – whether that is to those with purpose-driven trips to the area (as we did on the way to Ten-ei village) or to those passing through (see diagram).

The flow is simple: these michi no eki provide parking, and a small shop or food stand. The site showcases local specialties (yams and rice at Ten-ei) or educate about the area (now we know what a yacón is and how much is grown in Ten-ei; it can also serve as a delicious substitute for many root-based dishes!). Then, the calculus for effectiveness for smaller michi no eki such as Kinosato Ten-ei is how much of the nearby expressway’s traffic it can divert.

For many others, the amenities arms race is one strategy – make the station a destination in and of itself. For Kinosato Ten-ei it may be promising for the michi no eki to act as an innovation hub for testing out regional revitalization strategies devised by the area’s residents. As volunteer teachers we tasked our pupils in Ten-ei village with this endeavor and heard many proposals brimming with energy and regional pride – from guided ghost tours to value-added products. There is still a lot to explore about michi no eki, but the flexibility of programming these spaces leaves many doors open.

Resources

I used several GIS shapefiles to put together the maps in this blog post. You can access them (and do your own mapping!) here:

I also used a modified Marker icon created by Fatkhul Karim for The Noun Project to demarcate the michi no eki.

Categories
Internal migration Japan Urban analytics

Moving to the beat: Seasonal trends in Japan’s internal migration data

Kendrick Leong

March 14, 2020 saw the earliest declaration of cherry blossom season in Tokyo on record. But while this symbolic start of spring may fluctuate year to year – a week earlier in 2020, and from March 21 in 2021 – other spring starts pepper the Japanese calendar in March and April. The start of the academic school year and corporate onboarding of new recruits accompany the blossoming of sakura trees. And data from the Report on Internal Migration in Japan, published by the Ministry of Internal Affairs and Communications’ Statistics Bureau, suggests that these starts may be reflected in internal migration patterns.

Internal migration is calculated by the Statistics Bureau by tracking legally required reports to municipal officials whenever a Japan resident moves between municipalities. What stands out is the spike in internal migrants – including those moving between prefectures as well as those moving within prefectures – in March and April of 2020. Our first question here, as with countless other exploratory data analyses about 2020, is naturally, but unfortunately: Is this pandemic related?

Let us reference that the first COVID-19 case reported in Japan was in January; timelines may line up here. Could we be seeing the rumored urban exodus debunked elsewhere? To find out, we bring in data for the past five years, starting in 2016.

What we see in the five-year data is similar spikes in March and April, consistently, every year. And these spikes are large – nearly 1.5 million Japanese residents move between municipalities in these spring months. Let us also reference that the first state of emergency in response to COVID-19 was declared on April 7, 2020, for Tokyo and later expanded to the entire nation. This state of emergency lasted until May 25, 2020, bookending the spring migration spikes.

The strength of the spring migration spike may have weathered the state of emergency. In April 2020, nearly 376,000 residents moved within their prefecture, and another 388,000 residents moved to a new prefecture, for a total of around 764,000 internal migrants nationwide. However, it is apparent that the state of emergency heavily impacted internal migration for May of 2020.

Internal migration tanked for May 2020, recording a forty percent decrease over 2019 numbers, and coming out to about 120,000 fewer individuals moving across municipalities. Some of these moves may have been postponed to June, with the lifting of the state of emergency. However, internal migration continued to fall short of 2019 numbers for the summer months and into the early autumn. A small uptick in November 2020, the latest month for which data is available, could represent moves that were postponed until the winter.

We could entertain any number of explanatory hypotheses: fewer work-related moves, most interestingly those of “business bachelors,” often young males that are dispatched to distant corporate posts; the move of some university semesters online; and the growing popularity of remote work (although not yet widely adopted). We hope to further investigate some of these topics in future blog posts.

Working with the Report on Internal Migration in Japan data, we can further explore some of the trends within the larger decline in migration nationwide in 2020. Inter-prefectural migrations – that is, across prefecture lines – show a decrease in April 2020 compared to a year earlier. This may suggest some reluctance for big moves amidst the national state of emergency.

Both migrations across and within prefectures mirror larger trends in internal migration for Japan as a whole. In intra-prefectural migrations – those within prefecture lines – we see a growth over 2019 numbers in April. This increase in year-on-year intra-prefectural migrations, combined with April decreases in inter-prefectural migrations, creates a wash on the national level, compared to 2019. In general, the migration decreases seen over the summer and autumn are less pronounced amongst migrations within the prefecture.

As we tinker with the Report on Internal Migration data, we can revisit some of the points brought up earlier. The March and April 2020 migration spikes are cyclical – we have seen these spikes yearly in the five-year data. As far as the nationwide state of emergency goes, it did not significantly impact the March or April migration spikes; however, its effects can be seen in the May drop, as well as consistent year-on-year decreases through the summer and autumn. It may very well be that the signs of spring in Japan – cherry blossoms, college entrance ceremonies, corporate initiations, and now, as we see, internal migration – still held in 2020.

In the next part of this series on migration we will zoom in on more granular data from the Report on Internal Migration. We have looked at when people are moving, we will next turn to where people are moving. We will also begin to examine migration trends specifically for the Tokyo metropolitan area.