loop of ufunc does not support argument 0 of type float which has no callable log method

[mouzui]

PyLandStats version: 2.4.2
Python version: 3.12.3
Operating System: MacOS
Description
I encountered an issue while using the conditional_entropy function in the pylandstats package. The error message suggests there is a problem with the compute_entropy function when it attempts to handle the log method on a float. I was expecting the function to compute the conditional entropy without any errors.

What I Did
I ran the code:
This resulted in the following traceback:

Traceback (most recent call last):
File "path/to/your/script.py", line 153, in
cauculate(tif_file)
File "path/to/your/script.py", line 105, in cauculate
CONDENT = lscape.conditional_entropy()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "path/to/pylandstats/landscape.py", line 2638, in conditional_entropy
return self.joint_entropy(base=base) - self.entropy(base=base)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "path/to/pylandstats/landscape.py", line 2611, in joint_entropy
return compute_entropy(adjacencies, base=base)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "path/to/pylandstats/landscape.py", line 116, in compute_entropy
entropy = -np.sum(pcounts * np.log(pcounts))
^^^^^^^^^^^^^^^
TypeError: loop of ufunc does not support argument 0 of type float which has no callable log method
Proposed Solution
To fix this issue, I suggest modifying the compute_entropy function to ensure that counts is converted to a NumPy array with a float data type before performing any operations. Adding the following line at the beginning of the function should resolve the problem:

python
Copy code
def compute_entropy(counts, base=None):
counts = np.asarray(counts, dtype=float)
pcounts = (counts / counts.sum())[counts > 0]
print("Proportional counts:", pcounts)
entropy = -np.sum(pcounts * np.log(pcounts))
if base:
entropy /= np.log(base)
return entropy
This change ensures that the counts variable is correctly handled as a NumPy array of floats, preventing the TypeError related to the log method.

Thank you for your attention to this matter.

Best regards

[martibosch]

Hello! would it be possible to have access to the data that raised this error? thanks!

[mouzui]

Hello! would it be possible to have access to the data that raised this error? thanks!

Sure.
https://1drv.ms/u/s!AuHP1_XRM7dihMcNnI_rK3ma3LK4Rw?e=nudt3t