What We Are Reading

The Future of Jobs 2018 by World Economic Forum - Link

World Economic Forum has published a long report about how the future of jobs will be impacted by AI. It’s pretty much the same story we have been hearing. The repetitive jobs will be replaced by AI, jobs with Data Science skills will be in more demands, etc.

But one chart I thought was interesting to share is this one.

It shows the top and bottom 10 job types based on their demands over the last 5 years based on LinkedIn data.

It turned out the top job in North America is Real State Agent, which doesn’t have much to do with AI or data science. But I’m not personally surprised after having seen this crazy boom in the real state market in Bay Area (San Francisco / Silicon Valley). I’ve even heard that the ones who have benefited the most in this super economic growth backed by the tech industry are the landlords who lease offices to those tech startups. ;)

Anyway, this goes to show that there are always type of works that don’t look as popular from the outlook but will be in high demands.

How statisticians are trying to change the way we measure poverty - Link

Royal Statistical Society in the UK, which is responsible for managing the stats in the country, has recently redefined what poverty means. The new metric rule takes not only the income but also the asset into account. Many pensioners with low income but high assets will be taken out from the poverty counts.

Changing the definition of the metrics like this can have a big impact at the downstream. In this case, some electoral districts or demographics might realize the poverty issue more than before. This would make a new set of people including politicians all the sudden realize this issue as an urgent issue. And this will form different policies around the poverty.

Data analysis can help you understand why the poverty is getting worse or better, but the most important thing is to define what the poverty is. This is why maintaining (or updating) the business metrics periodically is critical for any businesses. For example, startups like us have different metrics to monitor as we grow our businesses.

Underneath all the AI hype is the likelihood it threatens the poor - Link

AI will widen the inequality of wealth further.

Another concern about AI, this time it’s from Kai-Fu Lee, a former Google China president and an investor mainly in AI.

I have talked about Data Network in this Weekly Update before. The more data you have the better product and service you can build, which makes more people to use, hence the first mover advantage. So AI and Big Data should help the incumbents, not the new kids on the block.

But looking at something like Tesla and Uber / Lyft, it’s not the incumbents (e.g. Toyota, Ford, Yellow Cab, etc.), it’s the new kids, who know how to take advantage of AI and Big Data to provide better services, end up threatening the incumbents.

And this means there are huge opportunities for new companies to compete against big companies. When was the last time a new car company threatened the incumbents in Japan and Germany?

In this era, we all have to constantly learn and change. The ones who refuse to change will end up being swept by the next wave. I tend to think this is rather exciting.

Concerns about AI Monopoly - Link

Meredith Whittaker, a Google/New York University AI researcher, expresses her concerns about the AI monopoly by a few tech companies that are mainly run by “white, male, affluent, and located in the Bay Area.”

Every time I go visit Japan I see more people use the services from Bay Area such as Google, Apple, Twitter, Instagram, Netflix, etc. And they are becoming more dependent on the “AI recommendation” by such services to make their daily decisions. Can AI built by the Bay Area monopolies serve Japanese well? Not sure.

But, if you know how media industry works, our thinking and opinions have been always formed largely by a handful media companies which have also been dominated by, ah, “white, male, affluent, but not located in the Bay Area” So yes, we should be concerned by such monopoly, but at the same time such a concern is not new and the alternative could be even worse. Algorithm with bias is something you can fix. Human with bias, that’s hard to fix.

First Notebook War - Link

“I feel a major difference between the R culture and Python culture is that Python users seem to create code more often, whereas R users often use code. There seems to be a strong atmosphere of software engineering in the Python world: in the beginning was the custom class (with methods). For R users, in the beginning was the data.”

It has been ongoing discussions about Jupyter Notebook’s shortcomings among data practitioners for the last few weeks.

Originally, it started by Joel Grus when he presented “I Don’t Like Notebooks” to claim that “Data Scientists should be more serious about programming, and if you are serious about programming you should not use Jupyter Notebook.”

But it also opened up arguments about Software development vs. Data Science, Python vs. R, the style of the programming, etc. among many data scientists

Data Scientists tend to downplay people who use UI tools for their lack of programming skills. Then, Software developers tend to downplay people who use Notebooks for their not adequate programming skills.

To me, this type of argument is just a waste of time. It’s not about the tool, it’s about what problems you are trying to solve with data.

No single solution fits all. Use the tools that help you solve your problems the most efficient way. And I’m seeing most of the data analysis problems don’t need programming, or at least as long as you have Exploratory. ;)

What We Are Writing

This week, we have written the following post.

• Creating R environment where R script from Exploratory can run https://exploratory.io/note/hideaki/6838376453273360

What We Are Working On

Getting closer, but we still have many issues to address before releasing v5.0. We have done this many times, but it never gets easy.

Anyway, as I have told you last week, we are introducing a big UI/UX change in v5.0. One of such changes is about Chart “Pinning”.

This feature is useful if you know how it works, but the problem is, well, it was not as obvious! especially, when you are just starting off the journey with Exploratory.

So we have made a few changes in this area.

First, the chart “pinning” is the default now, which means any new chart you will create will be ’Pinned" to the currently selected step. But this also means that you will want to switch the “Pinned” step often. So we have made it easier to switch the “Pinned” step with drag-and-drop.

Another enhancement in Chart area, you will be able to just scroll the chart tabs when you have many charts and see the thumbnail images to see what those tabs are.

No more ‘More’ tab, if you know what I mean. ;)

Data Science Booster Training

Our next online Data Science Booster training will be in this coming November. If you are interested in learning Data Science without programming, make sure to sign up soon!

Enroll November Booster Training!

If you are a current student, click here to get the student discount.

That’s it for this week.

Have a wonderful week!

Kan CEO/Exploratory

Subscribe Weekly Updates!

This is a weekly email update of what I have seen in Data Science / AI and thought were interesting, plus what Team Exploratory is working on.