Difference between revisions of "Hierarchical Clustering"
| Line 15: | Line 15: | ||
The code uses the AgglomerativeClustering from sklearn.cluster, which doesn't have GPU support. | The code uses the AgglomerativeClustering from sklearn.cluster, which doesn't have GPU support. | ||
| + | |||
| + | If this is being run on a new build box then | ||
| + | pip install statistics | ||
| + | pip install gmplot | ||
The input is a tdt file named CoLevelForCircles.txt with 7 columns: | The input is a tdt file named CoLevelForCircles.txt with 7 columns: | ||
| Line 20: | Line 24: | ||
The output is a tdt file named Results.tsv with 8 columns: | The output is a tdt file named Results.tsv with 8 columns: | ||
| − | (city, state, year) layer cluster ('lat','long','coname','datefirstinv') | + | (city, state, year) layer cluster ('lat','long','coname','datefirstinv') |
==Documentation== | ==Documentation== | ||
Revision as of 14:20, 29 May 2019
| Hierarchical Clustering | |
|---|---|
| Project Information | |
| Project Title | Hierarchical Clustering |
| Owner | Kyran Adams, Oliver Chang |
| Start Date | 12/1/2017 |
| Deadline | |
| Keywords | Cluster, Clustering, Circles, Pain in the ass, Agglomeration |
| Primary Billing | |
| Notes | |
| Has project status | Active |
| Copyright © 2016 edegan.com. All Rights Reserved. | |
Summary
The code is in
E:\projects\hca
The python3 file is main.py
The code uses the AgglomerativeClustering from sklearn.cluster, which doesn't have GPU support.
If this is being run on a new build box then
pip install statistics pip install gmplot
The input is a tdt file named CoLevelForCircles.txt with 7 columns:
city state year lat lon coname datefirstinv
The output is a tdt file named Results.tsv with 8 columns:
(city, state, year) layer cluster ('lat','long','coname','datefirstinv')
Documentation
There's useful reference material here: https://stackabuse.com/hierarchical-clustering-with-python-and-scikit-learn/
Note that it should be possible to use Tensorflow's KMeansClustering to achieve the same result.
Old Code Notes
This code takes a CoLevel master file, clusters points using k (number of clusters) in the range [1, num points / 5), and creates a file output.tsv.
Output.tsv has columns place, statecode, year, layer, cluster, lat, long, coname, datefirstinv. Layer is k, and cluster is the id of the cluster that the point belongs to.
The original version by Kyran and Oliver is in:
E:\McNair\Projects\FastCircles\src
You can run this program with:
python3 main.py
