Jupyter Notebook¶
Manipulate Notebooks from the Command Line¶
See StackOverflow.
Execute notebook and save results
jupyter nbconvert --execute --to notebook --inplace notebook.ipynb
Export notebook to
.pyjupyter nbconvert --to python notebook.ipynb
Export notebook to
.html(Convert without execution)
jupyter nbconvert --to html notebook.ipynb
(Execute before conversion, with no execution time limit)
jupyter nbconvert --execute --to html --ExecutePreprocessor.timeout=-1 notebook.ipynb
(Execute before conversion and set output dir to
./html)jupyter nbconvert --execute --to html --ExecutePreprocessor.timeout=-1 --output-dir='./html' notebook.ipynb
Programmatically Create Jupyter Notebooks (with different parameters)¶
Specify parameters in the first cell of the source jupyter notebook (source.ipynb):
You can then use nbparameterise to programmatically replace input values in a notebook before running it.
import os
import nbformat
from nbparameterise import extract_parameters, parameter_values, replace_definitions
# Extract parameters in source notebook
with open("source.ipynb") as f:
nb = nbformat.read(f, as_version=4)
orig_parameters = extract_parameters(nb)
# Parameters to replace
model_names = [
'model3-small-search-space4', 'model3-small-search-space5'
]
# Create new notebooks with replaced parameters
for name in model_names:
print("Running for model:", name)
# Update the parameters
params = parameter_values(orig_parameters, model_name=name)
new_nb = replace_definitions(nb, params, execute=False)
# Save
new_nb_fp = f"new_notebook_with_param_{name}.ipynb"
with open(new_nb_fp, 'w') as f:
nbformat.write(new_nb, f)
# Execute new notebook
os.system(f"jupyter nbconvert --execute --ExecutePreprocessor.timeout=-1 --to notebook --inplace {new_nb_fp}")
# Convert to html
os.system(f"jupyter nbconvert --output-dir='./html' --to html {new_nb_fp}")