3 ms·
What the heck is going on with that initial scatter plot? The Y-axis is log-base3, which is nonstandard but technically valid. The X-axis? log-base2 times $30k?
by minimaxir 11y ago
What the heck is going on with that initial scatter plot? The Y-axis is log-base3, which is nonstandard but technically valid. The X-axis? log-base2 times $30k? It doesn't work like that, and it might be misleading to portray SF/California as outliers, especially when different bases are being used for the axes.
Yes, SF real estate is expensive, no one will deny that. But the core argument appears to be "don't buy SF real estate because the economy is on a downturn" which is specific to neither real estate nor SF.
- ChuckMcM 11y agoIts a way to emphasize the point, and I agree with you that it pushes it into deception. That said, it seems like most of the people looking at homes in my neighborhood both people work, and generally at engineering median[1], not area median salaries. So when you're annual income is 250K plus, spending 6K a month on a $1.5M mortgage is 1/3 your income. [1] http://www.indeed.com/salary/q-Software-Engineer-l-Silicon-Valley,-CA.html http://www.indeed.com/salary/q-Software-Engineer-l-Silicon-V...
- GFK_of_xmaspast 11y agoHow is that any different than a log-log plot that's been squished in one axis? What I'm smdh about is that least squares fit, I want to see what happens if you use a Huber loss.
- ScottBurson 11y agoNo, the core argument is "don't buy SF real estate because interest rates are going up". It's a valid point. Given a certain amount of income, a home buyer can afford a certain monthly payment. If interest rates go up, and prices didn't fall, payments would go up. Since the market can't support higher payments -- people are paying as large a fraction of their income as they can stand to already -- what will happen instead is that prices will have to come down. As for the plot, it's just a log-log plot. Stretch the axes however you like; it's still going to have the same shape.
- lintiness 11y agoif the author knows which way interest rates are going, he should quit his day job and buy / short interest rate futures. the road is littered with bodies of people who believe japan's interest rates couldn't stay at "zero" forever.
- cowsandmilk 11y ago> when different bases are being used for the axes. um... there is a constant for converting between any two log bases... so I have no clue what you are claiming is wrong. It is a factor of 1.5849625007211563 for converting between log base 2 and log base 3... A log scale is a log scale is a log scale. There is no "log-base 3" or "log-base 2" scale.
- minimaxir 11y agoWould that factor still be valid since the X-Axis "log" is already being transformed by multiplying by 30k? That's what's really confusing me. (It might be so but that just makes interpretation way confusing and the article doesn't elaborate on that.) The Y axis (price to income) is derived from the X axis (income) as well, which will be affected by representing with different bases. I have no issues with log/logs for general visualization, but the use of it here seems like brute-force outlier identification which isn't representative of the typical relationship of California and other states. (Especially with a trendline)