Real-Time Air Quality Monitor

Summary 📑

• Prerequisites
• Setup
  • Apache Zookeeper, Apache Kafka, Logstash
  • Docker Compose
  • Importing a Dashboard in Kibana (UI)
  • Useful links
• Set Real-Time data with crontab
  • Start manual data ingestion
• Getting new training data
  • Update your model
• Screenshot

Prerequisites 📜

To use the Air-Quality Monitor app, you should have familiarity with the following technologies:

• OpenWeather (Register to obtain an API KEY)
• Docker
• Docker Compose
• Python
• Crontab Docs

Setup ⚙️

To set up the Air Quality Monitor, follow these steps:

Apache Zookeeper, Apache Kafka, Logstash

Download Kafka by running the following command:

cd kafka/setup
wget https://downloads.apache.org/kafka/3.7.2/kafka_2.13-3.7.2.tgz
cd ..

Edit the version if necessary Versions

template@version-edit:$~ wget https://downloads.apache.org/kafka/[NEW_VERSION]/kafka_[NEW_VERSION]

Docker Compose

Run the following command to start the Docker Compose:

docker compose up --build

Importing a Dashboard in Kibana (UI)

To import the dashboard on Kibana you must:

1. Open the left-side panel and click on Stack Management.
2. Navigate to Saved Objects.
3. Click Import (top-right corner) and select the file /kibana/kibana_dashboard.ndjson

Useful links:

• Kafka UI
• Spark UI
• Elasticsearch UI
• Kibana UI

Set Real-Time data with crontab

After running docker-compose, the ingestion_manager container will complete its work in a few minutes. To generate additional data, you can set up an automatic job with crontab for other containers that handle the ingestion_manager.

• Open the crontab editor with crontab -e
• Add the following line to the crontab: 0 * * * * cd /full/path/air-quality-monitor && /usr/bin/docker compose up ingestion_manager >> /full/path/air-quality-monitor/cron.log 2>&1
• Save the crontab file

Start manual data ingestion

To start a manually real-time version of the app with real values, run the following command:

docker run -it --rm --hostname="ingestion_manager" --network aqm -e DATA_ACTION="NODEMO" air-quality-monitor-ingestion_manager

Getting new training data

Just like the real-time data, you can also collect historical data to train your model.

• Check the following link to see the available data: Historical Data
• Generally use the API call: http://api.openweathermap.org/data/2.5/air_pollution/history?lat={lat}&lon={lon}&start={start}&end={end}&appid={API key}
• Save the historical data in the data folder with a name according to the load() function on the save_old_data.py file. Actually it is milan_3months.json

with open('../data/milan_3munths.json') as f:
    data_raw = json.load(f)

• Run python3 save_old_data.pyW

Update your model

1. Uncomment the train_model service in the docker-compose.yml file.
2. Build its image with docker compose build train_model.
3. Start the container with docker compose up train_model.
4. Check the model's files in /spark/model