This Repository is for Spatiotemporal Innovation Workshop Held by Spatial Data Lab , Center for Geographic Analysis, Harvard Univeristy
More Detail :Spatiotemporal Innovation Summer Workshop 2024
Harvard DataVerse : https://dataverse.harvard.edu/dataverse/sdl_dataverse
KNIME Geospatial Extension : https://github.com/spatial-data-lab/knime-geospatial-extension
The workshop mainly use KNIME Analytic Platform and its Python Extension.
-
Go to the KNIME website (choose suitable version of Operation System)
https://www.knime.com/downloads/download-knime
Note:as the updating of KNIME is frequent, self-extrating archive is highly recommended for Windows**
-
Extract to a folder, double click KNIME.exe to start KNIME
-
To get the best performance of KNIME, open knime.ini with Notepad(Windows)
revise -Xmx2048m as -Xmx8g ( 8 Giga Byte, or half for your computer memory)
-
Choose a directory as workspace(e.g. E:\KNIMEworkspace),While starting KNIME
user can execute multiple KNIME APP simultaneously (same or different versions), but each KNIME Analysis Platform must use a unique directory as workspace;
While run only one KNIME APP at a time, different versions of KNIME software can use the same workspace directory.
-
Go to Anaconda website and download
Double click the installation file to run Anaconda
-
Star Anaconda Prompt or Anaconda,build a new Conda enviroment of Python 3.9 for KNIME
conda create -n geoai python=3.9 knime-extension knime-python-base -c conda-forge -c knime
-
Python packages
2-2-GeoAI
pip install XGeoML mgwr geoshapley matplotlib pygam
pip install git+https://github.com/hyperopt/hyperopt-sklearn
remember to install git first
3-1-Network Community
pip install fiona python-igraph leidenalg pygeoda
pip install torch torch_geometric
refer to: https://pytorch-geometric.readthedocs.io/en/latest/install/installation.html
3-2-Remote Sensing
pip install rasterio folium mapclassify
pip install earthengine-api geemap
-
Install Python and R extension in KNIME
open KNIME > File > Install KNIME extension , in the dialog
input "Python" in the box of type filter text bar > check KNIME Python Integration > input "R Statistics" in the box of type filter text bar > check KNIME Interactive R Statistics Integration > next >... > Restart KNIME
-
Commomly used Extensions
Geospatial Analytics Extension for KNIME
KNIME H2O Machine Learning Integration
KNIME XGBoost Integration
KNIME Al Assistant (Labs)
KNIME Al Extension
KNlME Deep Learning - Keras Integration
KNIME H20 Machine Learning Integration
KNlME Machine Learning Interpretability Extension
KNIME XGBoost Integration
KNIME Column Expressions (Labs)
KNIME Plotly
KNIME Data Generation
KNlME Textprocessing Chinese Language Pack
*KNIME Email Processing
*KNIME Nodes for Scikit-Learn (sklearn) Algorithm
*KNIME Parallel Chunk Loop Nodes
*KNIME Statistics Nodes (Labs)
*KNIME Web Interaction (Labs)
*KNlME Interactive R Statistics Integration
*Redfield NLP Nodes
KNIME > Preferences > KNIME > Python Deep Learning > click "New environment"
- Configuring R and Python enviroment in KNIME
open KNIME > File > Preference, in the dialog
KNIME > Conda > click Browse... to choose the installed Anaconada directory, such as : D:\ProgramData\Anaconda3
KNIME > Python > choose geoai(or other environment) for Python3 (defult)
KNIME >R >click Browse... to choose the installed R directory, such as C:\Program Files\R\R-4.2.1
click Apply and Close.
-
Go to CRAN website to download R package
choose suitable version of Operation System
e.g. R 4.2.1 for Windows : https://cran.r-project.org/bin/windows/base/
Recommend : Download R for Windows > Install R for the first time > Previous releases> R 4.1.3 (March, 2022)
-
RStudio Installation (RStudio is not necessary)
Go to RStudio website
https://www.rstudio.com/products/rstudio/download/
https://download1.rstudio.org/desktop/windows/RStudio-2022.07.1-554.exe
Double click to install it.
-
Install Rserve package
Rserve is the basic pacakge to run R extension
for KNIME R 4.2.x might need to install its source data directly.https://www.rforge.net/Rserve/files/
the best way to install Rserve without installing RStudio
install.packages('Rserve', dependencies = TRUE)