Model Details
Model Description
- Developed by: Warsaw Student Hacking Team
- Model type: Multi
- Language(s) (NLP): Pytorch
- License: agpl-3.0
Prediction of 10 PM10 emissions in Berlin based on traffic intensity measured by 80 stations across city (more about stations here -> https://api.viz.berlin.de/daten/verkehrsdetektion). Emissions train data extracted from here -> https://www.umweltbundesamt.de/en/data/air/air-data/stations. Model uses traffic station's num_vehicles, quality, hour, month concatenated in this order.
Used traffic monitor stations detid_15:
[100101010073424, 100101010075343, 100101010075444, 100101010075545, 100101010073323, 100101010077161, 100101010072717, 100101010072616, 100101010035331, 100101010043617, 100101010043516, 100101010085750, 100101010055741, 100101010055640, 100101010079585, 100101010066047, 100101010085649, 100101010069885, 100101010069986, 100101010002086, 100101010053923, 100101010029570, 100101010054024, 100101010029469, 100101010059983, 100101010002692, 100101010074838, 100101010074939, 100101010061603, 100101010061704, 100101010018355, 100101010018456, 100101010067259, 100101010017547, 100101010017648, 100101010067158, 100101010042708, 100101010042809, 100101010076656, 100101010076555, 100101010077060, 100101010076959, 100101010045132, 100101010045233, 100101010062512, 100101010062411, 100101010062613, 100101010062714, 100101010060084, 100101010085952, 100101010040179, 100101010040078, 100101010073525, 100101010073626, 100101010002288, 100101010083427, 100101010083528, 100101010053014, 100101010027348, 100101010013709, 100101010023914, 100101010083629, 100101010013810, 100101010024116, 100101010002389, 100101010024217, 100101010024419, 100101010053115, 100101010024318, 100101010035230, 100101010079787, 100101010027247, 100101010079080, 100101010078979, 100101010074232, 100101010074131, 100101010072212, 100101010072111, 100101010023510, 100101010023611]
Used PM10 monitor stations codes (for model training):
DEBE032, DEBE061, DEBE051, DEBE056, DEBE065, DEBE069, DEBE010, DEBE034, DEBE063, DEBE068
Evaluation
Evaluated on randomly choosen subset of prepared data.
Metrics
Mean absolute error used in validation.