Simple Transformer Temperature Forecasting

What am I?

This repository contains the example code talked about in this application note, presenting a method of using the Disruptive Technologies (DT) Wireless Temperature Sensors for forecasting future values by applying simple signal modelling. Written in Python 3, it uses the DT Developer API to communicate with a DT Studio project and its sensors. By calling sensor_stream.py, a temperature forecast will continuously be calculated for previous history data and/or a live stream of datapoints from the moment of execution.

Before Running Any code

A DT Studio project containing temperature sensors should be made. All temperature sensors in the project will attemped forecasted upon. More information regarding project setup can be found in the application note.

Environment Setup

Dependencies can be installed using pip.

pip3 install -r requirements.txt

Edit sensor_stream.py to provide the following authentication details of your project. Information about setting up your project for API authentication can be found in this streaming API guide.

USERNAME   = "SERVICE_ACCOUNT_KEY"       # this is the key
PASSWORD   = "SERVICE_ACCOUT_SECRET"     # this is the secret
PROJECT_ID = "PROJECT_ID"                # this is the project id

Usage

Running python3 sensor_stream.py will start streaming data from all sensors in your project for which a forecast be calculated for either historic data using --starttime flag, a stream, or both. Provide the --plot flag to visualise the results.

usage: sensor_stream.py [-h] [--path] [--starttime] [--endtime] [--plot]
                        [--plot-init] [--plot-debug]

Temperature forecast on Stream and Event History.

optional arguments:
  -h, --help    show this help message and exit
  --path        Absolute path to local .csv file.
  --starttime   Event history UTC starttime [YYYY-MM-DDTHH:MM:SSZ].
  --endtime     Event history UTC endtime [YYYY-MM-DDTHH:MM:SSZ].
  --plot        Plot the resulting forecast.
  --plot-debug  Plot algorithm operation.

It is essential that the user inspects the properties of their data and configure the parameters in ./config/parameters.py accordingly as no one configuration can work for all data.

Note: When using the --starttime argument for a date far back in time, if many sensors exist in the project, the paging process might take several minutes.