deap | Distributed Evolutionary Algorithms in Python | Machine Learning library

There are no limitations to the use of SHAP for model explanations as they are literally meant to

explain the output of any machine learning model

(compare with the docs)

There are indeed several core explainers available. But they are optimized for different kinds of models. Since your case consists of a CNN model built with TensorFlow and Keras, you might want to consider the following two as your primary options:

• DeepExplainer: meant to approximate SHAP values for deep learning models.

• GradientExplainer: explains a model using expected gradients (an extension of integrated gradients).

Both of these core explainers are meant to be used with Deep Learning models, in particular such built in TensorFlow and Keras. The difference between them is how they approximate the SHAP values internally (you can read more about the underlying methods by following the respective link). Because of this, they will very likely not return the exact same result. However, it is fair to assume that there won't be any dramatic differences (although there is no guarantee).

While the former two are the primary options, you might still want to check the other core explainers as well. For example, the KernelExplainer can be used on any model and thus, could also be an option on Deep Learning models. But as mentioned earlier, the former two are particularly meant for Deep Learning models and should therefore (probably) be preferred.

Since you are using pictures as input, you might find the image_plot useful. Usage examples can be found on the GitHub repository you already linked to. You also do not need to worry about colors as they do not matter (see the DeepExplainer MNIST example).

Both strategies can possibly select the same individual multiple times, so replacement is not the point.

I don't know how np.random.choice is implemented internally, but anyway the implementation method is not specified in the contract of the function (and so it could change at any time). Below I give my implementation of both selection strategies using numpy. Please, make sure to test these functions before using them in anything serious.

EDIT: sorting by fitness was not necessary; I don't know what I had in mind.

You could use the Python built-in function getattr, which lets you get attributes of an object by name.

As explained in the docs: getattr(x, 'foobar') is equivalent to x.foobar.

So you could do something like:

It's because you're declaring deap inside your if statements:

I've got some experience with rasterio, but I am not nearly a master with it. If I remember correctly, rasterio requires you to have installed the program GDAL(both binaries and python utilities), and some other dependencies listed on the PyPi page. I don't use conda at the moment, I like to use the regular python 3.8 installer with pip. Given what I'm seeing with your installation, I would uninstall rasterio and follow a different installation procedure.

I follow the instructions listed here: https://rasterio.readthedocs.io/en/latest/installation.html
This page also has separate instructions for those using Anaconda.

The GDAL installation is by far the most annoying but once it's done, the hard part is over. The python utilities for both rasterio and gdal can be found here:
https://www.lfd.uci.edu/~gohlke/pythonlibs/#gdal
The second link is also provided on the PyPi page but I like to keep it bookmarked because there's a lot of good resources there!

I cannot reproduce your error, check if the following code works for you:

I don't think DEAP package can do that. You can simply implement the algorithm on your own or find the new packages.

Anyway, check out my library contains most of the state-of-the-art meta-heuristic algorithms. It also evaluates the entire population in each generation. https://github.com/thieunguyen5991/mealpy

I now know what happend here. When this question was originally asked, the image in question did not have the make command inside it.

Adding RUN apt-get install -y build-essential to the Dockerfile solves the issue. But, this image is built on dockerhub, so when the image was rebuilt after adding this instruction - it was not refreshed locally. You would actively need to pull the image again like I mention later in this answer.

I verified this, but rebuilding the image from your original Dockerfile definition and confirming that make is not in that image.

I checked, and the image haipengzhang/merchant_score_project does infact have have the make command in the path /usr/bin/make, and it is executable. The user in the image is root, and I am able to start the make command just fine.

There is a little problem with your docker run command, but I do not think that it is related to the issue:

You should not preceed the $(pwd) command with a / in the volume mapping. Try this instead:

Pycel compiles multi-sheet Excel spreadsheets easily. Pycel's function pycel.ExcelCompiler is lazy and so doesn't actually compile Excel into Python until explicitly asked to do so via excel.evaluate('Result!E16')} (i.e. evaluate some cell)

I tried: