The Electric Rice Cooker

An NHL Playoffs Race Simulator at La Presse

In North American ice hockey, guessing which are the team matchups for the postseason can often seem arcane. Not anymore with the playoffs race simulator we did at La Presse.

We started work on the simulator about 10 days ago. I had found a great sports stats website called Sports Club Stats while reading tweets from one of The Montreal Gazette’s Habs reporters. The site has a section where it lists the chances your team will finish at a particular standing rank, which is important for first-round playoffs matchups.

A site with numbers is nice, but it’s probably even more interesting if you could see the possibilities for yourself.

To develop the Simulator we used D3.js, a great library for building applications in the browser where the document (the “webpage”) needs to change according to the data that you load and then interactively change.

We had in mind to develop the Simulator for the iPad, thus some of the design choices like an interface dominated with clickable shapes and buttons, rather than scroll-down menus and text boxes.

Such an interface could possibly be refurbished to be a “calendar visualiser”. It’s D3.js, so you could resize the axes as you please, although we didn’t judge it necessary to do so in this iteration of the application. The horizontal axis is the date and the vertical is the current rank, but there is no reason why you couldn’t line up the games by their score differential, or re-order the teams by their number of overtime wins, etc.

Finally, the fun factor is the most important. Why a hockey schedule simulator? It gets crucial at the end of the season, as the number of games left and amount of standing points remaining to be distributed become easier to calculate mentally. Easier to calculate, but still real hard to visualise (unless you have a big sheet of paper, a calculator and a game schedule nearby).

You can calculate all the scenarios, with tie-breakers and complex mutually exclusive situations taken into account by the simulator so that you can avoid having a headache. Games between teams is the second tie-breaker and quite complex to program right, because you need to know the entire schedule and how many away and home games have been played.

This application is a first in the hockey world, and perhaps in the north american sports scene, as far as I know. You can justify its development cost because hockey is a religion in Quebec and in the rest of Canada.

To top it off, you give it the title for what most people in La Presse’s home market will use it for: figure out who their Habs will play against in the first round of the Stanley Cup playoffs. But you might as well use it for any of the team you prefer following.

Archived the original Rice Cooker blog for JMSC HKU

A bit of housekeeping: the blog that I wrote in between 2010 and 2012 (formerly at jmsc.hku.hk/blogs/ricecooker/) has been archived to my own domain: http://cedric.sam.name/ricecooker/. The new stuff post-2012 has been and will continue to be located here on this Tumblr (electricricecooker).

2 months ago

Reuters’ Connected China webapp (really an iPad app)

Reuters came out with an impressive webapp on China and its politics called Connected China. It is divided in five main sections: China 101, Social Power, Institutional Power, Career Comparison and Featured Stories.

http://connectedchina.reuters.com/

The real power of this webapp is the wealth of information. For sure, when you land on the page, you might be wondering what you are looking at. The app itself lacks permalinks linking to specific states (the hashmarks you usually find on dynamically loaded pages like Gmail).

But if you do make the dive inside the data, you will find absolutely everything you wanted to know about political power in China — and even a dose of encyclopaedia-like articles on the recent history of China.

The career comparison is a really neat visualisation. Think of the man-hours needed to compile and curate this data!

In a country where political information is opaque, distilling and visualising the data (for human consumption) remains the best way to make predictions. But then again, who knew Obama would become president before 2006?

Now I wonder how Reuters is going to integrate this reference in its China coverage. Will they create permalinks to link their articles to particular parts of the webapp? Will they write stories based on the data compiled for this project?

One year of deleted weibos archive

We started the systematic search of deleted post just a year ago, sometime in the first week of February 2012. Since then, we found tens of thousands of weibos disappeared by Sina censors…

3 months ago

An overview of the Hong Kong budget in 2012-13

datajournalismlab:

This is for the fiscal year currently being completed. The Hong Kong government will announce numbers for 2013-14, and we’ll try to update the numbers. This visualisation is made to highlight the difference of magnitude in spending for different departments in Hong Kong.

Source: datajournalismlab

The Data Journalism Lab Blog: Southern Weekend affair: WeiboScope detects an exceptionally high volume of deleted posts on Sina Weibo

The WeiboScope at JMSC wasn’t designed to provide an accurate measure of the volume of deleted posts. It samples some users’ timelines (latest 100 posts or so) at definite times, usually a few times per day, and compares copies to find deleted ones.

However, we have a handful of people whose posts we constantly check, at every few minutes. So, if that person makes a post, then it is immediately caught in the archive running behind WeiboScope. When a post is deleted, we get an e-mail alert right away.

The last few days have been rather exceptional. I get about an alert per month usually, but the Southern Weekend affair has generated a few e-mails per day since it exploded on January 3rd/4th.

Here is a sample of posts caught on the fly, which is by no means scientifically chosen:

Want to know what gets censored on Sina Weibo in real-time? We have an automated tool, and a public website:
http://research.jmsc.hku.hk/social/search.py/sinaweibo/#lastpermissiondenied

Since the Chinese characters for “Southern Weekend” can no longer be searched, WeiboScope has become an invaluable tool to identify posts that are written about the scandal.

Not everything gets deleted, and these entries are just what we manually found through our home-brewed WeiboScope trending topics:

http://research.jmsc.hku.hk/social/index.py/singleSinaWeibo?id=3530828434765336
http://research.jmsc.hku.hk/social/index.py/singleSinaWeibo?id=3530900413205911
http://research.jmsc.hku.hk/social/index.py/singleSinaWeibo?id=3530902254434129
http://research.jmsc.hku.hk/social/index.py/singleSinaWeibo?id=3531053517742307
http://research.jmsc.hku.hk/social/index.py/singleSinaWeibo?id=3531064133928981
http://research.jmsc.hku.hk/social/index.py/singleSinaWeibo?id=3531123780610767

E-mail alerts since last summer

It is difficult to give an exact quantification of the extent of censorship happening. The alerts are also tailored towards the users we “super-track” (at every few minutes, instead of only a few times per day).

However, it is for sure the most e-mail alerts I’ve received in a few days since the height of the Bo Xilai / Wang Lijun story in March to May 2012 (we started the deleted posts tracking in February 2012).

The story is still evolving and our colleagues at China Media Project are keeping their eyes and ears on it.

Source: research.jmsc.hku.hk

The Data Journalism Lab Blog: An overview of the Hong Kong budget in 2012-13

The 2013-14 Hong Kong budget will soon be released, in about thee weeks.

In the meanwhile, we have this visualization for 2012-13:
http://opengov.jmsc.hku.hk/hongkong/budget/2012/

Where does the government spend most of their money? Consistently with other places in the world like the USA: social security, health, and education.

Obviously, it’s nowhere near finished, and not so useful now. But at least, it serves the purpose of giving an overview of relative amounts spent per department and budget area.

(via datajournalismlab)

Source: opengov.jmsc.hku.hk

The Data Journalism Lab Blog: It's almost Hong Kong Budget season! Here's some data!

When I first moved to Hong Kong, these ads were running on television about the upcoming Hong Kong budget, starring finance secretary John Tsang:

Harry Harrison, the South China Morning Post cartoonist, had a different view of them:

Harry's View on 2010-11 Tsang budget in SCMP

Whereas many Western countries struggle to balance their budgets, Hong Kong is notorious for its humongous surpluses (even compared with projected figures) and cash handouts.

The budget in Hong Kong is announced in an address by the Financial Secretary (still John Tsang, despite a new CE and ExCo) every February 1st.

We scraped the numbers from PDFs (such as this one for 2012) released on the budget website every year and are publishing them in a Google Fusion Tables:

https://www.google.com/fusiontables/data?docid=1PLEJLAGgchvfHIG277KU_4gFVJTsEviWwOXiOhY

The years are for the numbers in each of the budget document. So for “2012”, the actual expenditure is for 2010-11, and the approved and revised estimates are for 2011-12. Only the estimate is for 2012-13.

Consistently high up there are the Social Welfare Department, the Government Secretariat: Education and Manpower Bureau and the Government Secretariat: Health, Welfare and Food Bureau (Health and Welfare Branch).

Source: datajournalismlab

“Animated Transitions in Statistical Data Graphics”, by Heer and Robertson at Berkeley / Microsoft Research. So it makes sense that someone would have studied this, especially with the advent of visualization-focused libraries like D3.js, from which I found this study. (That said, this study dates back to 2007.)

Source: vis.berkeley.edu

Hong Kong Air Pollution Index Explorer (alpha)

Hong Kong Air Pollution Index Explorer

After extracting a bunch of data from the Environmental Protection Department’s website on air pollution, I proceeded to make a visualisation:

http://opengov.jmsc.hku.hk/hongkong/airpollution/indexexplorer/

The data continually updates itself from data scraped from EPD, and the visualisation follows the stream and changes every time new hourly data is release.

We designed the “Hong Kong Air Pollution Index Explorer” as a media product that lets you compare the current API with past data. Because it just makes better sense this way.

This is released in alpha, because I literally just finished it and have not yet seriously debugged. I’ll post on the Data Journalism Lab website when it’s more done.

You can try the “Sort” and “Colour” buttons to switch around the different views. This is the cool feature! Made with D3.js. :)

Source: opengov.jmsc.hku.hk