Open mohamedelshami opened 5 months ago
To install qlib from source,
first pull the code locally,
then install numpy and cython: python -m pip install "numpy<=1.23.5" cython
,
and finally install qlib using python setup.py install
.
I don't know if your procedure is the same as the one I described, but I hope my answer will help you.
Thank you, I've tried your suggested instructions. I keep running into various library conflict errors which I'm not able to resolve.
Python 3.8.17
RuntimeError: Scikit-learn requires Python 3.9 or later. The current Python version is 3.8.17 installed in
Python 3.9.13 error: The 'pymongo==3.7.2' distribution was not found and is required by pyqlib
Not sure what system you are using? I tried installing it under ubuntu and windows and found no conflicts. Currently qlib support system include windows
and linux
.
I tried both on Windows and Mac with conda environment
Try the following:
conda create -n qlib python=3.9 numpy pandas matplotlib pyyaml hdf5 requests -y
conda activate qlib
conda install - c conda-forge lightgbm liblapack
just support python <= 3.9
I've made few attempts to install pyqlib using both pip install and build from source.
I use conda environments and tried with 3 different Python versions 3.8.15, 3.8.17 and 3.9.13.
The common issues encountered are:
pip install pyqlib
Defaulting to user installation because normal site-packages is not writeable ERROR: Could not find a version that satisfies the requirement pyqlib (from versions: none) ERROR: No matching distribution found for pyqlib
Same error when trying Jupyter Notes with Google Collab: https://github.com/microsoft/qlib/blob/main/examples/workflow_by_code.ipynb
This
3.9
ERROR:: Could not find a local HDF5 installation. You may need to explicitly state where your local HDF5 headers and library can be found by setting the
HDF5_DIR
environment3.8 RuntimeError: Scikit-learn requires Python 3.9 or later. The current Python version is 3.8.15 installed
Any pointers would be great.