Neural network to predict particulate levels for the next 7 days

Hi,
Using Purpleair outdoor sensors, we built a neural network with the aim of predicting pm10 and pm2.5 levels for the next 7 days based on weather forecasts.
We have released the first version of the neural network in opensource format, on which we have made some improvements at a later time.
The repository of what we have made available in opensource format can be reached at this on github searching the project: “APP-Air-Pollution-Prediction”.
On the company website you can see the results of the prediction made by the neural network: https://www.pignataropatrimonio.it/pm10-pm2-5-forecast/.
An intern from the University of Naples Federico II collaborated in the realization of the project, who summarized the work done in this book (in English): https://www.amazon.it/Application-forecast-Particulate-Learning-algorithms/dp/B09QNZRFSR/ref=sr_1_1?__mk_it_IT=ÅMÅŽÕÑ&crid=1ZKHLVZKNE0ZE&keywords=Web+Application+to+forecast+airbone+Particulate+Matter+levels+with+Deep+Learning+algorithms&qid=1651219579&sprefix=web+application+to+forecast+airbone+particulate+matter+levels+with+deep+learning+algorithms%2Caps%2C71&sr=8-1.
We are planning a second phase to extend the analysis to the data of a network of sensors scattered throughout the territory (at the moment we have 4 operational and one under installation) and to obtain the forecast at one hour intervals over the course of the day.

To tell the truth, I agree mostly with the aim of finding a little enthusiasm to find the crica to continue in this project that I hoped my fellow citizens would appreciate who instead simply ignore (the project is carried out at no cost to citizenship).
So thank you everyone for listening to this little rant.

3 Likes

very interesting, congrats!
surely there are many people interested to this approach, maybe they are outside Pignataro now, then they will come too…
I follow you

1 Like