GitHub - georgedashen/DSAA6000G

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 31 Commits
area_feature		area_feature
data		data
evalulation		evalulation
final_project		final_project
model		model
test_week		test_week
README		README
combined.py		combined.py
config.py		config.py
lgb_binary.py		lgb_binary.py
lgb_magnitude.py		lgb_magnitude.py
mergeData.py		mergeData.py
mergeFeature.py		mergeFeature.py
pred.py		pred.py
readData.py		readData.py
toolkit.py		< 5A49 div class="overflow-hidden"> toolkit.py

Repository files navigation

This is for AETA earthquake prediction project of course DSAA6000G, midterm

All scripts are written via Python3.9.2. 

Proper folder structure has been created for directly running scripts under the data_and_model folder.

toolkit.py is the same as the original one provide in baseline.

config.py has been modified to include all features, not just abs_mean, from EM and GA files.

mergeData.py is the same as the original one provide in baseline.

readData.py has been modified to generate engineered features for EM and GA separately because the feature matrix is too large.

MergeFeature.py is a new script to combined EM and GA data for each area and separated as train and valid.

lgb_magnitude.py trains the multi-class model.

lgb_binary.py trains the binary model.

combined.py makes the final prediction for validation set using the two models.

pred.py makes prediction for both magnitude and location of test data.

area_feature folder contains data for training and validation.

data folder contains earthquake events information and station informatino.

model folder contains the two models for each area.

evaluation folder contains performance matrix for the three models.

test_week folder contains EM/GA data and prediction from a 7-day window.