Una serie de acotaciones al margen a medida que voy leyendo algunos libros... A series of annotations whilst reading interesting books... A collection of notes on books about science, SciFi, history, others topics... Una colección de notas sobre libros de ciencia, ciencia-ficción, historia, otros...

Showing posts with label Rant Raves. Show all posts
Showing posts with label Rant Raves. Show all posts

16 July 2011

Getting the indexes right–Midiendo los índices correctamente

Statistics can have a bad reputation including for the lay person, but it is a field with incredible real life applications. You have to be very skilled to use it and very gifted to elaborated on it. But most important very carefully to read and interpret the results. Maybe the last statement is the most important for all of us, being key decision makers or voters.

Indexes or KPIs in the MBA jargon are important to have an idea where you are and what corrected measures to take. It is used for everything but everyday is taking more importance for defining and deciding which are the best policies, from school systems, health care reforms, or just guiding the economy of a country, last but not least important. The debate is always heated as in today US health care reform, and reading carefully the data is important in order to keep the debate reasonable and not fall in partisan views.

Comparison between health care cost and life expectancy – source National Geographic

This is why I was sorry reading about the meddling of the actual Argentinean government with the Inflation Index or Consumer Price Index (CPI): Lies and Argentine statistics with the subtitle of “Stalinist practices In Buenos Aires”.

“MOST Argentines reacted with a shrug when their government began doctoring its consumer-price index in 2007. Cooking the books cost holders of the country’s inflation-linked bonds at least $2.3 billion last year (…) They reckon that inflation is now running at about 25%. That is far above the 10% reported by INDEC, the government statistics agency, but less than the 30% wage increases public employees have received in recent years.”

“A presidential election looms in October and inflation, and the government’s denial of it, is perhaps the biggest threat to the prospect of President Cristina Fernández winning a second term. That may be why Guillermo Moreno, the thuggish commerce secretary, is moving to stamp out the unofficial, but widely trusted, price indices. To do so he has dusted off a decree, penalising misleading advertising, approved by a military dictatorship in 1983. In February he sent letters to 12 economists and consultants ordering them to reveal their methodology, on the grounds that erroneous figures could mislead consumers. Some of Mr Moreno’s targets refused; the rest were analysed by INDEC, which predictably found their methods flawed”

And going back to statistics and the power of big numbers I found an article in the Sciam.com titled Economists Find Faster, Cheaper Way to Measure Inflation.

“Even in the information age, the U.S. Bureau of Labor Statistics still gathers much of its data the old-fashioned way. Workers make phone calls to find out what dentists charge for pulling teeth, and they visit stores to write down the prices of CDs and Russet potatoes. In the end, the data are accurate but take a month or so to compile and analyze.

To speed things up, Alberto Cavallo and Rigoberto Rigobon, economists at the Massachusetts Institute of Technology, created the Billion Prices Project (bpp.mit.edu). Software indexes Web sites to track prices of more than five million goods from 70-plus countries and spits out inflation rates in real time”

The method behind the calculation is impressive and it yields on top of having a massive amount of prices (retail) today available on-line from a lot of countries (70).

At the moment of this writing they are collecting data and estimating the indexes, but worth checking regularly…

28 January 2010

Grande Apophis!

Report says scientists lack funds to meet Congressional goal for finding smaller "near-Earth asteroids"

Dificil conseguir presupuesto para mapear los NEO, no digamos embarcarnos en algun proyecto para desviar algún asteroide apuntado a hacer algún desastre como el de Yucatán (menores impactos ya son suficientemente malos, por no decir que uno bastante chico puede ser peor que el terremoto de Haití y el Tusanmi de Indonesia combinados...)

Sin embargo la probabilidad de que nos pegue alguno es 1 cada 30 mil años... Si un terremoto cada 200 años no convence a un gobierno de exigir reglas de construcción antisísmicas (cuando tenés que priorizar pobreza, hambre, violencia y los corruptos de turno), veo más difícil de convencer a otro o a varios de que se pongan las pilas con este tema...