Observe

This software is used to perform statistical tests on astrology.
It is organized around the notions of studies, commands and outputs.

Note: here, "planet" is used in its etymological meaning (= "moving body"), so the sun and the moon are also called planets.

Draft under construction
What are we doing ?

Data

We are processing dates, for example the data sent by Didier Castille: Screenshot of a00.csv, sent by Didier Castille
It contains 591 936 lines with 3 or 4 dates:
  • Mother birth day
  • Father birth day
  • Child birth day
  • Wedding date
(321 838 lines contain wedding dates)
Mother Father Child Wedding schema

Planets

For each date, we compute planetary positions. These are the ecliptic longitudes of the planets at a given date, expressed in degrees, between 0 and 360.
Astrological chart of Grothendieck 
+-----+------------+
| day | 1928-03-28 |
+-----+------------+
| SO  | 7.577      |
| MO  | 97.618     |
| ME  | 340.585    |
| VE  | 342.739    |
| MA  | 322.279    |
| JU  | 14.344     |
| SA  | 259.157    |
| UR  | 3.626      |
| NE  | 146.828    |
| PL  | 103.775    |
| NN  | 73.017     |
+-----+------------+

Here are the main planets:
Planet Sun Moon Mercury Venus Mars Jupiter Saturn Uranus Neptune Pluto
IAA code SO MO ME VE MA JU SA UR NE PL
Symbol ☼ ☾ ☿ ♀ ♂ ♃ ♄ ♅ ♆ ♇

Distributions

Then we compute distributions:
Distribution where each planet is represented individually
1
Distribution example: planets in Gauquelin sectors
2

For example, we take each planet separately and represent their positions in a single drawing (example 1).
Then we group the positions by packets (36 packets of 10° in example 2).
This grouping is called a distribution.
This gives one distribution per planet.
Gauquelin 1955, example with 12 packets

Control groups and expected ditributions

The distributions computed to represent our original dataset are called the observed distributions.
To see if these distributions show anomalies, we must compare them to the expected distributions.
We build control groups by randomly shuffling the data and compute their distributions. Observed distributions, control groups, expected distributions
Several control groups are built, and the mean (average) distributions are computed.
This is supposed to produce what we should observe in the "chance hypothesis".

Statistical tests

Once the expected distributions are computed, it is possible to use statistical tests to see if the comparison between observed and expected distributions show anomalies.
The first test is the chi-square test, which indicates if the difference between observed and expected distributions is significant.
If a difference is significant, the effect size is computed to know how much the anomaly affects the individuals of the group.
Next: Single date distributions

About

Observe is a CLI (Command Line Interface), a tool used in a terminal (console) which pemits to issue commands to work on the data.
For example: 
php run-observe.php death-fr chi2 full
Means: "for study death-fr, compute chi2 values of the split full"

Program started in december 2020 by Thierry Graff to compute a00 distributions for Nick Kollerstrom, who studied if these data show statistical anomalies.
Rewrite in 2026 to test Deaths in France since 1970.

Program developed and tested under Linux (Debian 13) with php 8.5. A priori, it should also work under Windows and Macintosh.