« Late-to-the-gate depression brought to you by the Census | Main | Making charts beautiful without adding unneeded bits »


Jon Peltier

In the first set of housing starts charts, for the South data, it is not clear which curve in the charts is 2008 and which is 2009. Seems to me 2009 should be the flatter one both unadjusted and adjusted, especially in light of the steep drop off during 2008 shown in the second set of charts.

In the second set of housing starts charts, I can see the same qualitative drop from 2005 to 2008 in all regions, even in the Northeast. Having three series in the same space, at the lowest part of the scale, masks the individual behavior.

Can you share any insight into the models for removing seasonal effects? Presumably a moving average is the major component, but is this too simple an approach?

Joe Mako

After looking at the second set of charts with a line for each region, I was prompted to look into the data, and produced:

After seeing differences between my chart and his, I emailed Kaiser Fung, author of the blog post, and he suspects that he may have miss-processed the data, but is unable to look into the details until this weekend.


Joe: Thanks for spotting the error. Just want to clarify: the only issue is the first chart in the second set of charts that shows unadjusted data over time. I will update that one when I get a chance.

Jon: You're also right that I was referring to the 2008 lines rather than 2009. Will fix that too when I have a chance.

The comments to this entry are closed.


Link to Principal Analytics Prep

See our curriculum, instructors. Apply.
Kaiser Fung. Business analytics and data visualization expert. Author and Speaker.
Visit my website. Follow my Twitter. See my articles at Daily Beast, 538, HBR.

See my Youtube and Flickr.

Book Blog

Link to junkcharts

Graphics design by Amanda Lee

The Read

Keep in Touch

follow me on Twitter