Aside from a few outages when the feed I was scraping changed format or disappeared, that was over 2 years of data. I have been contacted by probably about 10 researchers who wanted to use my data in their research, and wanted to know if any more complete data was available. (young-0.com/airquality actually surpassed my Harry Potter page as the most popular in my domain)
There wasn't before, but now there is. Last year the full history finally got put online officially, not just for Beijing but also for Chengdu, Guangzhou, Shanghai, and Shenyang. Thank you, US Embassy!
Since the whole point of this was to get data to analyze it, it seems a shame not to look at it. Recently the Chinese government has proudly announced things have gotten better. Have they? I don't trust the government announcement or their statistics (testing stations have moved around, and some are in remote, cleaner areas of Beijing). The US embassy figures should be more unbiased - not perfect, since they have a single testing station in a single location, so it cannot give results for Beijing as a whole. Also, I'm missing some readings, and of course it is subject to the unintentional bias that the calibration may be off. Still, the relative readings over time should be comparable.
Because the statistics take up a lot of room I'll put them on another page, and just put the summary charts on this page. These were made with all the data I have, from March 2009 to the end of February 2015. The data was taken from online, put in a Libreoffice spreadsheet, manipulated using emacs, and the charts drawn with Libreoffice.
And the end results of the analysis? The Chinese claims appear to have a basis. The average aqi has dropped and the number of blue sky days has increased. There is still a long way to go, but we are on the way.
The first chart simply shows the average, high, and lows for each month, similar to what is shown on a daily basis in the website's summary graph:
There seem to be annual peaks around the beginning of the year, right around Spring Festival - but is this real, or an artifact of a few days of heavy fireworks?
The second chart shows the percentages of readings where the AQI is in the ranges 0-50, 51-100, 101-150, 151-200, 201-300, 301-400, 401-500, and over 500. The green and yellow section on the left represents the percentage of good days, and the black, grey, and dark purple on the right the number of bad days:
For numerical summaries see the second page