Category: 3 – SHELF it (All Categories are 1 – Read ASAP!, 2 – BUY it!, 3 – SHELF it, 4 – SOMEDAY it)
Comments: This is a Statistics textbook. So, it definitely isn’t for everyone. However, if you are interested in Statistics, it is an awesome book. Lays out concepts with a ton of clarity.
Top 3 Learnings:
1. Let’s say you are tossing a coin.
The law of averages says nothing about an increased likelihood of a tail after 4 rows of heads. Instead, it says that – as you keep increasing the number of tosses, the chance error as a percentage of tosses keeps going down.
2. Regressions ONLY deal with associations or correlations. An increase in x is associated with an increase in y.
3. A test of significance gets at the question of whether an observed difference is real (alternative hypothesis) or just a chance variation (null hypothesis). The observed significance level is the chance of getting a test statistic as extreme or more extreme than the observed one. The chance is computed on the basis that the null hypothesis is right. Small values of p are evidence against the null hypothesis – i.e. the observed difference is real.
Book Notes here.
Category: 2 – BUY it! (All Categories are 1 – Read ASAP!, 2 – BUY it!, 3 – SHELF it, 4 – SOMEDAY it)
Comments: I think I might have called this a “Priority 1” book if it wasn’t for business school. This was a very good refresher on how to think about predictions and data. As the ultimate data geek, Nate Silver does a very good job introducing us to the world of prediction and statistics.
Top 3 learnings:
1. Sometimes, predictions change the nature of the thing. If everyone is using an app that predicts highway x will have lesser traffic, everyone could end up on highway x.
2. Bayesian approach was to make a small prediction and keep improving on it. Probability was seen by Laplace and Bayes as a step toward progress. Bayes theorem is concerned with conditional probability. Think probabilistically. Require you to accept that your subjective representations of the world are not truth.
3. Terrorist attacks are similar to earthquakes – high uncertainty. However, when you plot frequency and destruction wrought by terrorist attacks on a double logarithmic scale, it is a straight line!
The broken windows theory was embraced in the US despite limited scientific evidence perhaps because it is easier for police to imprison a 16 year old with drugs than solve a difficult crime.
Israel has taken the opposite approach – it treats small acts of terror as normal but has worked hard to eliminate large threats. Israel’s power law distribution curve looks different from what you might expect – due to their strategic choices.
Book notes here.