home > kero
Python package kero
This is a python package initially intended for pre-processing in machine learning. More functionalities have been added, and more will be added. For a quick try, go here.
Update 2021 May 03. This package is no longer supported.
pip install kero
Resolve the dependencies with the following commands.
pip install pandas pip install opencv-python pip install numpy
An example of pre-processing
If you start learning machine learning, you will realise that a lot of time is spent on pre-processing the data, as much if not (probably) more than training the model itself. I aim to make this process simple.
Look at the following figure. Some data columns, such as third and fourth are not in numbers. In order to put them in machine learning model, we can binarize them. The column third can be either “gg” or “not”. By binarizing, we convert a value “gg” into Boolean 1 in column “third_gg”, and “not” into Boolean 0 in the same column (in this case, the column is nicely binary, i.e. if not “gg” then it is “not”).
Another example from image processing: creating ground truth images.
Hope you find this useful, and do let me know if you find anything that needs improvements.