All the data used in this study can be accessed from the data folder. The data folder has 3 subfolders i.e
- AirQuality
- Met
- MODIS
AirQuality AirQuality folder contains raw airquality data for each location inside the AirQo folder. Also generated files after preprocessing and analysis for each location are stored in the location folder.
Met Met folder contains raw meteorological data.
MODIS Modis folder contains raw greenness data stored in each locations folder
- air_quality_data_preprocesing_and_analysis_kireka_jinja_road_rubaga.ipynb
- air_quality_data_preprocesing_and_analysis.ipynb for (makerere,bugolobi, makindye,mulago, bukoto)
- air_quality_data_preprocesing_and_analysis_lubowa.ipynb
- air_quality_data_preprocesing_and_analysis_nsambya_usembasy.ipynb
The outputs from running the files above are stored in the respective folder for the location.
One of the outputs i.e merged_hourly_airquality_greenness_meteorological_dataset.csv contains the data that is used for training, evaluating models for the respective location.
File listed below contains the python code for preprocessing data to format neural network can ingest and for training and evaluating
models(SVR and Deep LSTM) models.
- air_quality_prediction_experimentation.ipynb
NB: all generated files are stored in the respective location folder inside the AirQuality/AirQo folder
File airquality_prediction_analysis_package_list.txt contains the packages in the environment used to preprocess ,train and evaluate models.
data_analysis_using_R folder contains the results and data used for generating calendarplots in R using the openair package.