ANN correlation models

Correlation models for several IMs developed via ANN

ANN models are available under /models directory.

Reference

Aristeidou, S., Shahnazaryan, D. and O’Reilly, G.J. (2024) ‘Correlation Models for Next-Generation Amplitude and Cumulative Intensity Measures using Artificial Neural Networks’, Under Review

Quick start without tensorflow models

1. Install requirements

pip3 install -r requirements.txt

2. Run a sample code using the tabulated empirical values (example1.py)

from correlation_models import aso2024_correlation_int, supported_ims, \
    supported_im_pairs


# Print supported IM names
supported_ims()

# Print supported IM pairs
supported_im_pairs()

# Example 1 - SA-FIV3
period1 = 1.05
period2 = 0.5
corr = aso2024_correlation_int("SA-FIV3", period1, period2)

print(f"Correlation coefficient between SA({period1}s) "
      f"and FIV3({period2}) is {corr:.3f}!")

# Example 2 - Sa_avg2-Ds575
period1 = 0.6
corr = aso2024_correlation_int("Sa_avg2-Ds575", period1)

print(f"Correlation coefficient between Sa_avg2({period1}s) "
      f"and Ds575 is {corr:.3f}!")

3. Run a sample code using the predictive models (example2.py)

Recommended approach

from correlation_models import aso2024_correlation


# Example 1 - SA-FIV3
period1 = 1.05
period2 = 0.5
corr = aso2024_correlation("SA-FIV3", period1, period2)

print(f"Correlation coefficient between SA({period1}s) "
      f"and FIV3({period2}) is {corr:.3f}!")

# Example 2 - Sa_avg2-Ds575
period1 = 0.6
corr = aso2024_correlation("Sa_avg2-Ds575", period1)

print(f"Correlation coefficient between Sa_avg2({period1}s) "
      f"and Ds575 is {corr:.3f}!")

Savvinos-Aristeidou / ANN_correlation_models