Newby question. Can you program and run ArcPy without having ArgGIS Pro or ArcGIS Desktop installed? Is it enough to have Python and Jupyter Notebook installed along with the ArcPy plugin?
With Pro... python, Jupyter, arcpy and Pro are all contained within a conda package distribution, so No.
With ArcMap, No. Arcpy is at many levels tied directly to one of the esri products, so you just can't run stuff on machines with neither installed.
According to the ArcGIS API for Python System requirements page lists only 1 requirement:
System requirements
In order to use version ArcGIS API for Python, you need Python 3.5 or later.
I wouldn't say that statement is incorrect, just incomplete. Installing the ArcGIS API for Python in a clean Miniconda installation gives a better idea of requirements (I created a new arcgis virtual environment😞
C:\Users\bixb0012\Miniconda3>Scripts\Conda.exe install -n arcgis -c esri arcgis
The following NEW packages will be INSTALLED:
arcgis: 1.3.0-py35hdae95b8_1 esri
bleach: 2.1.2-py35_0
certifi: 2017.11.5-py35h456c6ae_0
colorama: 0.3.9-py35h32a752f_0
decorator: 4.2.1-py35_0
entrypoints: 0.2.3-py35hb91ced9_2
html5lib: 1.0.1-py35h047fa9f_0
ipykernel: 4.8.0-py35_0
ipython: 6.2.1-py35h4a2ac14_1
ipython_genutils: 0.2.0-py35ha709e79_0
ipywidgets: 5.2.2-py35_1
jedi: 0.11.1-py35_0
jinja2: 2.10-py35hdf652bb_0
jsonschema: 2.6.0-py35h27d56d3_0
jupyter_client: 5.2.1-py35_0
jupyter_core: 4.4.0-py35h629ba7f_0
markupsafe: 1.0-py35hc253e08_1
mistune: 0.8.3-py35_0
nbconvert: 5.3.1-py35h98d6c46_0
nbformat: 4.4.0-py35h908c9d9_0
notebook: 5.3.1-py35_0
numpy: 1.10.1-py35_0
pandas: 0.20.3-py35he2ce742_2
pandoc: 1.19.2.1-hb2460c7_1
pandocfilters: 1.4.2-py35h978f723_1
parso: 0.1.1-py35he39c48a_0
pickleshare: 0.7.4-py35h2f9f535_0
pip: 9.0.1-py35h691316f_4
prompt_toolkit: 1.0.15-py35h89c7cb4_0
pygments: 2.2.0-py35h24c0941_0
python: 3.5.3-5 esri
python-dateutil: 2.6.1-py35h6b299a3_1
pytz: 2017.3-py35hb294b1a_0
pyzmq: 16.0.3-py35h1861011_0
send2trash: 1.4.2-py35_0
setuptools: 38.4.0-py35_0
simplegeneric: 0.8.1-py35h6406082_0
six: 1.11.0-py35hc1da2df_1
testpath: 0.3.1-py35h06cf69e_0
tornado: 4.5.3-py35_0
traitlets: 4.3.2-py35h09b975b_0
vc: 14-h0510ff6_3
vs2015_runtime: 14.0.25420-0 esri
wcwidth: 0.1.7-py35h6e80d8a_0
webencodings: 0.5.1-py35h5d527fb_1
wheel: 0.30.0-py35h38a90bc_1
widgetsnbextension: 1.2.6-py35_3 esri
win_unicode_console: 0.5-py35h56988b5_0
wincertstore: 0.2-py35hfebbdb8_0
winkerberos: 0.6.0-py35_0
The ArcGIS API for Python (a.k.a. arcgis) has quite a few dependencies. Fortunately, if you are using Conda to install it, the dependencies are taken care of for you.
Note, installing arcgis through Conda also installs Python 3.5.3-5. The packaging of the ArcGIS API for Python has gotten tighter over time. Each release is now developed against a specific version of Python, which makes the statement, "you need Python 3.5 or later", on the requirements page partially incorrect.
If you install the ArcGIS API for Python on a machine without ArcGIS Pro, you will not have access to ArcPy, so some of the functionality of the ArcGIS API for Python will not be available.
Hi Joshua, thanks for your comment! You are correct, there are many other packages that will install alongside the API to give it additional functionality. ArcPy, which is only bundled with Pro, is one of these packages. The API can still accomplish a lot without ArcPy, it all depends on how you use the API.
You are right, that Systems Requirement page is a bit incomplete: we'll work to update it so it contains more information about the dependencies. Thanks for your input!