Anomaly Detection Live Demo Instructions

Please note: this live demo is only intended to demonstrate the Anomaly Detector API on any CSV file that follows a simple schema. This demo loops over the provided data (with fake timestamps) to demonstrate the anomaly detection API. It does not use the real timestamps provided in the raw data, and should not be used in any production scenario. Once the demo is running, you will be able to see the raw data and the results from the anomaly detection API in the browser. Alternatively, you can also use the online Anomaly Detector demo page to evaluate the API on your own data.

In your CSV file, you need at least two columns. A column for the timestamp and column for the values you want to perform anomaly detection on. In addition, you can include an optional dimension column that allows you to run the anomaly detection API across different categories (e.g. different sensors, different regions, or even different variables) through a drop down selection menu in the demo. The timestamp column should be in the ISO 8601 format YYYY-MM-DDTHH:MM:SSZ. If you intend to use your own CSV, please see the sensor_data.csv file for how to structure that CSV file.

Step 1 - Setup

The demo should work in any Linux-based environment using any integrated development environment (IDE). If you choose to use Azure Machine Learning (AML) compute, you should install VS Code locally to be able to run this demo. If you are not using AML, just skip ahead to Step 2.

With VS Code installed, navigate to your AML resource in the Azure portal, and click on Launch Studio. On the left, click on the Compute tab. Finally, under the Applications column, click the VS Code link corresponding to the compute instance that you would like to use. When asked by VS Code to Allow an extension to open this URI? click Open.

Step 2 - Creating the virtual environment and installing the dependencies

Open a terminal in VS Code, and run the following command to create and activate a new conda virtual environment and install the dependencies:

conda create --name anomaly-detection-demo python=3.7
source activate anomaly-detection-demo
pip install -r requirements.txt

Next, enter the following two lines in your terminal, replacing <anomaly-detector-api-key> and <anomaly-detector-endpoint> with the values from your Anomaly Detector resource. You can find them by navigating to your Anomaly Detector resource in the Azure portal, and copying the Anomaly Detector key and endpoint from the "Keys and endpoints" section of the resource details.

export ANOMALY_DETECTOR_API_KEY=<anomaly-detector-api-key>
export ANOMALY_DETECTOR_ENDPOINT=<anomaly-detector-endpoint>

Step 3 - Adjusting the configuration (Optional)

If you plan to use your own CSV to run this demo, take some time to review and adjust the default configuration in the demo.py file. For the purposes of this demo, the default configuration is stored within demo.py as a Python data class with default values. Feel free to modify the values to match your own needs.

    csv_name: str = "sensor_data.csv"  # Name of the csv file containing the data
    value_column: str = "sensor_readings"  # Name of the column containing the values
    timestamp_column: str = "timestamp"  # Name of the column containing the timestamps
    dimension_column: str = "sensor_name"  # (Optional) Name of the column containing a dimension (e.g. sensor name, or location, etc). If your data does not have this column, set it to None.
    window_size: int = 50  # Size of the window used to compute the anomaly score
    minute_resample: int = 5  # Resample the data to this minute resolution
    ad_mode: str = "entire"  # Anomaly detection mode to use. Can be "entire" for batch mode or "last" for last point mode.

If your data does not have the optional dimension column, please set the dimension_column in the demo.py script to None.

Step 4 - Running the demo

Finally, navigate to this directory in your terminal. Make sure you have the anomaly-detection-demo virtual environment activated, then run the following command:

bokeh serve --port 5599 --show demo.py

A browser tab should open, and the demo should start running. Any detected anomaly will show in red. If the browser tab doesn't open, try to open it manually by navigating to http://localhost:5599/demo in the browser. If the demo does not start running, please double-check that you followed the steps above correctly.

The demo will continue to run until you stop it. To stop the demo, close the browser tab and enter Ctrl+C in the terminal to stop the process.