The Orange3_Glueviz package
Orange add-on for calling Glueviz within Orange.
Please uncomment the lines in setup.py to install required libraries. As I'm not really sure of the earliest versions required, I just list the versions in use, and comment out it in case of breaking down your python package environment. Please decide which version to be installed on your need.
Required libraries:
# install_requires=["orange3>=3.24.1",
# "anyqt>=0.0.10",
# "sortedcollections>=1.1.2",
# "sip>=4.19.8",
# "glueviz>=0.15.2",
# "orange3-network>=1.5.0",
# "numpy>=1.17.4",
# "scipy>=1.3.2"],
Install:
python setup.py install
There's an example in ./example/example.py
If you need to pass data sets to Orange from Spyder, read the documents in ./doc and hack Orange as it is mentioned. The two files of Orange need to be modified is there already, just copy and past.
Mar,10th,2020
Add a widget named "network from tables" to connect Spyder igraph-python to orange3-network add on. Please refer to ./example for examples and ./docs for documents in detail.
First of all, call the function graph2tables() to transform an igraph into two orange table vertices & edges, then call orange and pass in the tables, and connect the "network from tables" widget to the two data table widgets to rebuild the graph.
There's a bug in orangecontrib.network.network.base.py, in the degrees() function of the DirectedEdges() class, you can copy the file in ./spyder-orange3 directly:
def degrees(self, *, weighted=False):
return self._compute_degrees(self.edges, weighted) \
+ self._compute_degrees(self.in_edges, weighted)
# the original line is wrong
# + self.compute_degrees(self.in_edges, weighted)
Mar,15th,2020
Add support of vertices & edges selection to the "Network From Tables" widget through a checkbox on widget GUI and a clean_network() function to verify vertices ids are in range, remove wrong edges and null vertices if it's necessary, reindex the ids of vertices if it's necessary. Please refer to the example workflow of airtraffic.net in ./docs.
If you select a subset of the vertices, the widget will check the "Remove nodes without any edge" check box automatically to force reindexing the ids of vertices, or you may get an index out of range exception when creating the network, as the id values may exceed the range of the underlying sparse array.
Mar,16th,2020
Address the issue of displaying vertices ids as real through setting the decimal of orange.data.variable.ContinuousVariable to 0, now it's integer as required.
# set the number of decimals to 0 for ids of vertices, source & target of edges
# to display as integer
self.vertices.domain.attributes[0].number_of_decimals = 0
self.edges.domain.attributes[0].number_of_decimals = 0
self.edges.domain.attributes[1].number_of_decimals = 0
Mar,18th,2020
Add a widget "Network To Tables" to transform a network into vertices & edges data tables. Add check to "Network From Tables" widget to check inputs and create an id column for vertices table if it's necessary. Update the icons for these two widgets.
Mar,21th,2020
Modify the "Network From Tables" widget to accept changes from the inputs, without deleting and creating a new widget again.
Modify the messages from the widget to give important Information, Warning and Error message icons on the canvas widget icon of Orange workflow.
Fix a bug of the Network From Tables widget in clean_network():
if removeNodes:
# nodes = set(list(edges_tb.X[:,0])) | set(list(edges_tb.X[:,1]))
nodes = set(list(X1[:,0])) | set(list(X1[:,1]))
Mar,22th,2020
Modify the "Network To Tables" widget to add important information icon upon widget icon in canvas workflow.
Add an example to construct a network with igraph-python from the scratch, then transform into vertices and edges tables, pass to Orange,read ./example/test-igraph1.py and python.ows for details.
There's an issue of igraph-python that labels wouldn't be shown in Spyder(IPython inline), it's mentioned here: igraph/python-igraph#185, and the workaround is here: https://stackoverflow.com/questions/30640489/issue-plotting-vertex-labels-using-igraph-in-ipython.
Mar,23th,2020
Modify the "Network From Tables" widget to remember original vertices and edges input, so that when vertices ids are reindexed and vertices or edges input is changed later, the widget can reference and recompute again, no need to delete and reconnect the other one's input to get the original data again.
May,22th,2020
Add modifications to Orange Geo, all is in the ./geo directory.
Notes:
-
Modification_to_orange_geo.txt, details of the modifications.
-
mapper.py and widgets/plotutils.py, widgets/owchoropleth.py, programs modified already. You should replace the access key of tile provider www.tianditu.gov.cn with your own in plotutils.py.
-
geojson/admin0.json, geojson/admin1-CHN.json, geojson/admin1-IND.json, the GeoJSON file modified already for use in China, you can replace those in Anaconda3\Lib\site-packages\orangecontrib\geo\geojson with them directly, and delete admin1-TWN.json, admin1-HKG.json, admin1-MAC.json at the same time.
-
You can download vector maps of China from https://www.gadm.org/download_country_v3.html, for admin2 and admin3 level, transform into geojson format and named them as admin2-CHN.json and admin3-CHN.json, then copy to the Anaconda3\Lib\site-packages\orangecontrib\geo\geojson directory for use.
-
If you want admin4 support of China, you can buy map data from third party such as http://www.dsac.cn/DataProduct/Index/2019, then named it as admin4-CHN.json, and copy it to the directory above.
-
The file ChinaMap.py is the source to check and modify admin0.json and admin1-CHN.json, admin1-IND.json for use in China.
-
The file AddressSeg.py is examples of geocoding with several major providers in China, you should replace the access key of provider www.tianditu.gov.cn with your own in it.
-
There're five examples of Orange Geo in the ./geo/example directory. covid-19.py prepares data for covid-19-CN-Cities.ows with admin0-2, and ChinaCountyExample.py prepares data for ChinaCountyExample.ows with admin0-3 and GuangDongTownsExample.ows with admin0-4. The covid-19-USA.ows is admin0-2 in animation. Data needed for the examples is in ./geo/example/data, you may need to modify the ows files for the change of data directory.
ChinaCountyExample.py also transforms the admin3-4 shp files of China into GeoJSON format Orange Geo required, and covid-19.py transforms admin2 shp file of China. Note that you need to buy the admin4-CHN.json data by yourself, it's illegal to publish it here without authorisation. As to the admin0 and admin1 files, they come from natrual earth at http://www.naturalearthdata.com/downloads/ , and is free to be modified and published.
The covid-19.ows example is a big example showing how to integrate all kinds of Orange widget components together, which is directly from three articles of the Orange team, and data is from https://github.com/CSSEGISandData/COVID-19 .
A. Data Mining COVID-19 Epidemics: Part 1 : https://orange.biolab.si/blog/2020/2020-04-02-covid-19-basic/
B. Data Mining COVID-19 Epidemics: Part 2 : https://orange.biolab.si/blog/2020/2020-04-09-covid-19-part-2/
C. Data Mining COVID-19 Epidemics: Part 3 : https://orange.biolab.si/blog/2020/2020-4-015-covid-19-part-3/
May,26th,2020
Add an example of customized business map for taxation data base on admin4-CHN.json of GuangDong province, at ./geo/example/ZhuHaiShuiWu.py.
May, 28th, 2020
Add an example of using geocode and tile maps from www.tianditu.gov.cn, functions called are in ./geo/TianDiTu.py , and usage is at ./geo/example/AddressShuiWu.py, workfolw file of Orange is ./geo/example/ZhuHaiShuiWuAddress.ows.
May,31th,2020
Merge vector map and vector markers of tianditu.gov.cn, modify related sources in ./contextily/tile.py, ./geo/widgets/plotutils.py, ./geo/example/ZhuHaiShuiWu.py, details are mentioned in ./geo/Modification_to_orange_geo.txt.
Jun, 1st, 2020
Modify ./geo/mapper.py to force loading map shapes every time init() runs, to avoid null pointer exception.
Jun, 16th, 2020
Modify ./geo/widgets/owchoropleth.py to add support to selecting color palette in the widget.
Jun,17th,2020
Modify ./Orange/widgets/utils/colorpalettes.py to add two custom red color palettes for China. Add ./geo/example/palette.py to create the palettes.
Jun, 20th, 2020
Modify ./geo/widgets/owchoropleth.py to add aggregated data to output for future usage by other widgets.
July, 2nd, 2021
Modify ./geo/contextily/tile.py, ./geo/TianDiTu.py, ./geo/widgets/plotutils.py to update http request headers for TianDiTu online map.
Jan, 7th, 2022
Modify ./geo/contextily/tile.py, ./geo/TianDiTu.py, ./geo/widgets/plotutils.py to update http request headers for TianDiTu online map.