pp2

This project provides a more robust and stable version of the Principal Path algorithm (1).

1. 'Finding Prinicpal Paths in Data Space', M.J.Ferrarotti, W.Rocchia, S.Decherchi

Downloads files (mandatory)

git clone https://github.com/erikagardini/pp2.git

Hot to use this code

You can install python requirements with

pip3 install -r requirements.txt

Run the experiments with the new proposed method

You can compute the Principal Path with the new proposed method as follow:

cd new_principalpath
python3 "name_of_the_experiment".py

where the parameter name_of_the_experiment can assume the following value:

• "2d_experiment": to reproduce the experiments with the 2d trivial data sets
• "face_olivetti_experiment": to reproduce the experiment with the olivetti data set
• "mnist_experiment": to reproduce the experiment with the mnist data set

The code produces the following output when the 2d experiment is run:

• the Dijkstra shortest path from the starting point to the ending point
• the adjasted path with points equally distributed along the shortest path
• the image of the path for different values of s
• the value of the evidence for each model, it allows to perform the model selection step

The code produces the following output when an high dimensional experiment is run:

• the starting figure
• the ending figure
• the Dijkstra shortest path from the starting point to the ending point
• the adjasted path with points equally distributed along the shortest path
• the resulting path for different values of s
• the nearest figures to each path waypoints for each model

Run the experiments with the original version of the method

You can compute the Principal Path with the original method as follow:

cd original_principalpath
python3 "name_of_the_experiment".py

where the parameter name_of_the_experiment can assume the following value:

• "2d_experiment": to reproduce the experiments with the 2d trivial data sets
• "face_olivetti_experiment": to reproduce the experiment with the olivetti data set
• "mnist_experiment": to reproduce the experiment with the mnist data set

The code produces the following output when the 2d experiment is run:

• the filtered data (if the prefiltering procedure is performed)
• the waypoints initialization
• the image of the path for different values of s
• the value of the evidence for each model, it allows to perform the model selection step

The code produces the following output when an high dimensional experiment is run:

• the starting figure
• the ending figure
• the resulting path for different values of s