Smart Citizen Data

Welcome to SmartCitizen Data. This is a data analysis framework for working with sensor data in different ways:

• Interacting with several sensors APIs
• Clean data, export and calculate metrics
• Model sensor data and calibrate sensors
• Generate data visualisations - matplotlib, plotly or uplot
• Generate analysis reports in html or pdf and upload them to Zenodo

Installation

You can check it out in the before installing if you want. Works with Python 3.10 or later (tested from 3.10 to 3.13).

You can just run:

pip install scdata

Plots (device.ts_plot(), maps, panels...) need the plotting libraries, an optional extra. Without them scdata loads and processes data, and plotting raises an error that says how to install them:

pip install "scdata[plot]"

Parquet needs the storage extra: the backups (local or S3: device.load_from_storage(), device.backup_to_storage()) and exports in parquet (device.export(..., file_format='parquet')):

pip install "scdata[storage]"

Examples (scripts and notebooks) are in examples.

Work on the source code

Simply clone the repository with:

git clone https://github.com/fablabbcn/smartcitizen-data.git
cd smartcitizen-data

Install scdata package with requirements:

python setup.py install

Or if you want to edit:

pip install --editable ".[dev]"

Tokens and config

Blueprints, calibrations and sensor names are read from BASE_POSTPROCESSING_URL, the same variable as smartcitizen-connector. By default it is Smart Citizen Flows, which manages this metadata.

A device of the Smart Citizen API is processed with a blueprint of its hardware (hardware_url in its postprocessing). A hardware lists its blueprints, at most one of each kind: process, long and backup. blueprint_kind selects the one to use, which also defines the sensors to load. Without it, the process blueprint, else the long one:

device = sc.Device(params=sc.APIParams(id=18154))
device.blueprint, device.blueprint_kind     # ('sc_air', 'process')
device.handler.blueprints                   # {'process': 'sc_air'}: the blueprints of its hardware

device = sc.Device(params=sc.APIParams(id=<device-id>), blueprint_kind='long')

The blueprint of the hardware is used before the one given (blueprint), unless override_hardware_blueprint=True. device.blueprint_kind is the kind of the blueprint the device uses. A device without a blueprint (no postprocessing, a hardware that is only backed up, or without a blueprint of the kind asked for) loads its sensors, but has no channels to process, and its blueprint and blueprint_kind are None.

Settings can be changed in ~/.config/scdata/config.yaml. The file only keeps what differs from the defaults (the defaults of new versions apply after upgrading), for instance:

data:
  reload_metadata: false
paths:
  export: /home/me/exports

The data folder is ~/.cache/scdata by default ($XDG_CACHE_HOME/scdata if it is set), with the raw, processed, export... folders inside. Each folder of paths is changed on its own, and only to a folder that exists: changing data does not move the others. To move them all, set XDG_CACHE_HOME.

Also, .env files will be picked from ~/.cache/scdata. An env.example is provided. If you want to upload data to Zenodo, you will need to fill set an environment variable called ZENODO_TOKEN in your environment. The .env file is not mandatory for scdata to work. It only serves for authentication if you want to POST or GET data without throttling. You can still do GET requests but they will be throttled if not authenticated/authorized by the platform. See user roles for more info.

Using with jupyter lab (optional)

It can also be used with jupyter lab or jupyter. For this install juypterlab.

Contribute

Issues and PR more than welcome!

Funding

This work has received funding from the European Union's Horizon 2020 research and innovation program under the grant agreement No. 689954