comparison README.txt @ 23:6891c5523b69

load with neighbors :)
author Jeff Hammel <k0scist@gmail.com>
date Sun, 25 Jun 2017 18:13:43 -0700
parents 811adc9736eb
children 40a9850076a4
comparison
equal deleted inserted replaced
22:e69cb496324e 23:6891c5523b69
71 a maximum number of neighbors of 2376 and a minimum value of 0 71 a maximum number of neighbors of 2376 and a minimum value of 0
72 with a mean of 222. This is stored in `neighbors.json` in `data`. 72 with a mean of 222. This is stored in `neighbors.json` in `data`.
73 There are 72897 cases with less than 100 in this box and 73 There are 72897 cases with less than 100 in this box and
74 18444 with less than 10. 74 18444 with less than 10.
75 75
76 Because of this, the closest 50 neighbors were
77 precomputed and cached as `neighbors.dat`:
78
79 ```
80 find-neighbors /home/jhammel/tensorflow/cities1000.txt -k 50 --latlon 1. 1. neighbors.dat
81 ```
82
76 83
77 ## Web Service 84 ## Web Service
78 85
79 A simple web application was developed using 86 A simple web application was developed using
80 `gunicorn` (http://gunicorn.org/) as the server. 87 `gunicorn` (http://gunicorn.org/) as the server.
99 106
100 - parallelism: the distance calculation is done serially. As such 107 - parallelism: the distance calculation is done serially. As such
101 it is a `O(10^10)` operation on the dataset. This should be improved 108 it is a `O(10^10)` operation on the dataset. This should be improved
102 and parallelized 109 and parallelized
103 110
111 - no through-the-web (TTW) testing was done except manual.
112 This should be corrected with Selenium and other headless testing
104 113
105 ## (Hopefully) Helpful Links 114 ## (Hopefully) Helpful Links
106 115
107 - http://geojsonlint.com/ 116 - http://geojsonlint.com/
108 - https://en.wikipedia.org/wiki/Haversine_formula 117 - https://en.wikipedia.org/wiki/Haversine_formula