Summary of Algorithms to Live By by Brian Christian and Tom Griffiths | Includes Analysis by Instaread

Author:Instaread [Instaread]
Language: eng
Format: epub
Publisher: Instaread
Published: 0101-01-01T00:00:00+00:00

Key Takeaway 7

Observation data with familiar distribution can be predicted using specific operations.

Analysis

According to Bayesian inference, having prior knowledge of a subject improves predictions about it, but that prior knowledge must be protected from bias. The different kinds of distributions of observation data suggest three different strategies for making predictions based on what can be observed in the past. Normal distribution, also called a bell curve, can be predicted by finding the data’s average. A power law distribution, like movie revenues, suggests multiplying the data by an average derived from other observations. Predictions from Erlang distributions involve adding an amount that is calculated as the average in other observations.

Familiar industries may contain submarkets that follow new distribution models, which means that a new prediction rule must be used. For example, in the modern publishing industry, the power law determines that some books will dwarf the majority of others in sales. However, the market for a single textbook would likely follow an Erlang distribution, with a reliable and repetitive number to approximate its sales year to year, until it is too outdated to use. If a publisher releases a set of books written by authors of similar fame and within the same genre, a normal distribution is more likely to take over, so that the publisher can predict one book’s performance by taking an average of the other books’ sales.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.