3 ms·
I'm not going to comment about the article without reading it carefully; but just in case there are any aspiring data analysts online there is a chart that is i
by roenxi 6y ago
I'm not going to comment about the article without reading it carefully; but just in case there are any aspiring data analysts online there is a chart that is interesting even without knowing anything about the data...
On page 64 (65 as the PDF numbers it), chart (b) "Distribution of savings rate" has an awkward little peak on the left hand side of the chart. It is very difficult to tell from inspection whether that is a real feature of the underlying data or if it is an artefact naive kernel density estimation.
If you ever see that pattern in your graphs, do consider trying logKDE [0] rather than ordinary Kernel Density Estimation a la geom_density() in ggplot. It also isn't perfect but the result is usually a bit better for things that approximate an exponential distribution.
[0] https://cran.r-project.org/web/packages/logKDE/vignettes/logKDE.pdf https://cran.r-project.org/web/packages/logKDE/vignettes/log...
- JamesBarney 6y agoI think that's a real effect. Intuitively I would imagine a lot more people save 0% (or close to 0%) of their income than save 50% of their income, I would also expect a lot more people to save or overspend 3% of their income than overspend or save 25% of their income.