Files of check-in [d581518913] in the top-level directory

Reconnect

List people I used to email but haven't emailed lately.

Usage

The program "bin/frommail" depends on nmh and is configured for my particular email setup. You probably need to rewrite it for yours.

The program "bin/toaddresses" depends on R and the package fpc. It reads the file created by "bin/frommail". You should not need to change this program.

From this directory, run "./reconnect". Candidates are listed in order of priority.

Method

Emails sent by me are grouped by the people I sent them to. Within each group, dates are clustered, and mean time between clusters is calculated. The time from the last cluster to the present date is treated as an draw from a Poisson distribution. The probability of this interval, or a more extreme interval, ... of this

Recipients are ignored when they have fewer than two clusters or when the p

Possible improvements

If I used different email addresses for the same person, it would be neat to consider them as one person. On the other hand, the present setup is neat in how it finds people's old email addresses from before people settled on serious names. (That might apply only to my generation.)

It would also be neat to mark addresses to be ignored. In most cases, when I don't write to someone for a while, it's because I don't want to.

Tips

Most of the addresses I don't want to write to are customer service of companies that I patronized. Consider stuff like this.