Available Metrics

The following pixel-wise metrics are available:

Categorical Metrics:

Metric	Description
`cm`	Confusion Matrix
`precision`	Positive Predictive Value (Precision)
`recall`	True Positive Rate (Recall)
`F1`	Harmonic Mean of Precision and Recall (F1 Score)
`accuracy`	Overall Accuracy
`csi`	Critical Success Index (CSI)
`far`	False Alarm Ratio (FAR)
`pod`	Probability of Detection (POD)
`gss`	Gilbert Skill Score (GSS), also known as Equitable Threat Score (ETS)
`hss`	Heidke Skill Score (HSS)
`pss`	Peirce Skill Score (PSS)
`sedi`	Symmetric Extremal Dependence Index (SEDI)

Continuous Metrics:

Metric	Description
`mae`	Mean Absolute Error (MAE)
`mse`	Mean Squared Error (MSE)
`rmse`	Root Mean Squared Error (RMSE)
`bias`	Frequency Bias
`drmse`	Debiased Root Mean Squared Error (DRMSE)
`corr`	Pearson Correlation Coefficient (Correlation)

Example Usage

Below are examples of how to calculate pixel-wise metrics using the provided functions:

from duplexity.pixelwise import calculate_pixelwise_metrics
import numpy as np

# Generate some random data to simulate observed and model output
observed_data = np.random.rand(100, 100)
model_output = np.random.rand(100, 100)

# Calculate pixel-wise metrics
results = calculate_pixelwise_metrics(observed_data, model_output, metrics=["mae", "rmse"])

print(results)

# Output:
# {'mae': 0.123, 'rmse': 0.345}

The calculate_pixelwise_metrics function allows you to compute multiple metrics at once by specifying them in the metrics argument.