The codes to construct random forests with missing data are found in https://github.com/IrvingGomez/Random_forests_with_missing_values. There, it is possible to find examples on its use, in particular the example Example_Anderson iris.ipynb shows most of the features.
Here are two images as an example, representing a decision tree with missing values.
This is a project in collaboration with Jill-Jênn Vie from Inria-Lille, France.
Autoencoders Dimensionality Reduction In some applications like data visulization, data storage or when the dimmensionality of our data is to large, we’d like to reduce its dimmensionality of the data, keeping as much information as possible. So we’d like to construct an encoder that takes the original data and transform it into a latent variable of lower dimmensionality.
Analysis of poverty in Mexico using public datasets and development of a poverty index using autoencoders.
Multivariate analysis of the data from the WHR 2017