pyFIMTDD (Legacy)

A implementation of the FIMT-DD regression-tree algorithm, based on the paper "Learning model trees from evolving data streams"

The implementation is not based on existing implementations. The version is debugged and tested for one-dimensional Data and Labels.

Files

pyFIMTDD.py is the actual FIMTDD-algorithm.
FIMTDD_LS.py an attempt to improve FIMTDD by replacing the leafnode-approximators with adaptive filters. Effectiveness ist still not proven. Current results are approximately 5-10% reduction in the cumulative Loss.
FIMTDD_evaluator.py an eval-framework for optimizing parameters and testing the algorithms

Name		Name	Last commit message	Last commit date
Latest commit History 27 Commits
.idea		.idea
__pycache__		__pycache__
.gitignore		.gitignore
ChristmasTreeLS.py		ChristmasTreeLS.py
DataGenerator.py		DataGenerator.py
FIMTDD_LS.py		FIMTDD_LS.py
FIMTDD_evaluator.py		FIMTDD_evaluator.py
Greedy_FIMTDD_LS.py		Greedy_FIMTDD_LS.py
KielClean.csv		KielClean.csv
Legendre_Test.py		Legendre_Test.py
README.md		README.md
abalone.data		abalone.data
flightdatascaler.pkl		flightdatascaler.pkl
large scale kiel.png		large scale kiel.png
optimal_performance.png		optimal_performance.png
optimzer.py		optimzer.py
preprocessor_script.py		preprocessor_script.py
pyFIMTDD.py		pyFIMTDD.py
result.txt		result.txt
results.txt		results.txt
scaler.pkl		scaler.pkl
surface3d_demo.py		surface3d_demo.py
ucFIMTDD_LS.py		ucFIMTDD_LS.py
umcFIMTDD_LS.py		umcFIMTDD_LS.py
wire3d_demo.py		wire3d_demo.py