This Python script calls https://opendata.lillemetropole.fr API to get air quality data from their "indice-qualite-de-lair" dataset and generates automated reports for each city.
├── README.md <- You are here.
├── data
│ ├── raw <- Extracted data gets uploaded here in the form of csv and pickle file
│ └── processed <- Clean data with air quality index and air pollutants details per city
│
├── notebooks <- Draft stuff
│ │
├── reports <- Generated html reports
│ └── figures <- Generated graphics and figures
│
├── requirements.txt <- The requirements file for reproducing the environment
│
└── src <- Source code for use in this project.
├── data
│ └── extract_data.py <- Script to extract the data from this 'very well-known site' and generate csv files in data/raw
│ └── transform_data.py <- Script to transform csv files from data/raw to interim and processed files
│
└── visualization
└── visualize.py <- Scripts to create visualizations
The data retrieved from "indice-qualite-de-lair" Lille Metropole dataset is actually originating from Atmo HDF.
It contains data from 1st January 2022 for the 95 cities of the European Metropolis of Lille. It is composed of a daily main indicator as well as 5 sub-indices, each representative of an air pollutant:
- Nitrogen dioxide (no2)
- Sulfur dioxide (so2)
- Ozone (o3)
- Particulate Matter of of less than 10mm (pm10)
- Particulate Matter of of less than 2.5mm (pm2-5)
The highest sub-index determines the index of the day. For example, if all indexes are at 1 (good) except one that is at 4 (Poor), the general index will be at 4.
Python and its various libraries:
pandasto manipulate datanumpyto work with arrays etc.matplotlib.pyplotto create and customize visualizations etc.
- The level of granularity of the data is quite poor. In an ideal world I would have had access to the raw data of the measurements instead of their sub-indexes. With the data I've worked with, I'm not able to differentiate a 51 µg/m3 Particulate Matter 10 from a 100 µg/m3 measurement for example. They would both fold into the "Poor" category despite being quite different from each others.
- I only get data from 01/01/2022 from this source of data
| O3 | SO2 | NO2 | PM10 | PM2.5 | Level |
|---|---|---|---|---|---|
| 0 to 50 | 0 to 100 | 0 to 40 | 0 to 20 | 0 to 10 | Good |
| 50 to 100 | 100 to 200 | 40 to 90 | 20 to 40 | 10 to 20 | Fair |
| 100 to 130 | 200 to 350 | 90 to 120 | 40 to 50 | 20 to 25 | Moderate |
| 130 to 240 | 350 to 500 | 120 to 230 | 50 to 100 | 25 to 50 | Poor |
| 240 to 380 | 500 to 750 | 230 to 340 | 100 to 150 | 50 to 75 | Very Poor |
| > 380 | > 750 | > 340 | > 150 | > 75 | Extremely Poor |
This project is open-source and available under the MIT License.