Skip to content

Latest commit

 

History

History
43 lines (26 loc) · 1.89 KB

TODO.md

File metadata and controls

43 lines (26 loc) · 1.89 KB

TODO

* ~~Refactor out matplotlib dependencies present in clusteringModels and predictiveModels
  modules, replacing with calls to the plotter module(which wraps bokeh and seaborn)~~

* Add a dask/airflow/luigi/pinball support for training models with different samples on
  distributed systems.

* Cleanup/refactor the plotter.py to remove obsolete/unused plots

* Add support for feature filtering..(tsfresh module and also others) in features.py

* Add Gini Coefficient-like measure visual for the cluster analyze

* Add support for https://github.com/ANNetGPGPU/ANNetGPGPU in the cluster analyze logic

* Add 3D heatmaps and may be 3D + 1D(time) visualizations/Animations(like the gapminder
  bubble chart for ex:)

	* analyze TODO: May be add a way to plot joint distributions of two variables?

	* ~~analyze TODO: add grouped violinplots by categorical variables too.~~

* Add a separate grid search function to grid search a data set with the given
  model.(wrapper around sklearn model_selection's grid search)

* Add Gaussian Mixture Model to clustering models

* Add factor_analyze function to analyze.py(probably something like PCA or the likes)

* Add plots for regression analysis with different models(may be [r-squared like]
  http://scikit-learn.org/stable/auto_examples/linear_model/plot_lasso_and_elasticnet.html#sphx-glr-auto-examples-linear-model-plot-lasso-and-elasticnet-py)
  or somethin else

* Add a way to check for non-linear correlations(aka ace algorithm)

* Implement the trellis plots for correlation analyze (when there's categories)

* Add support for [RANSAC](https://en.wikipedia.org/wiki/Random_sample_consensus)

* Setup python sphinx and add proper documentation for all classes and functions

* Create a function to take dataframe, run tree/randomforest, pick out best tree, create a
  neural network based on the tree, and return it.. (The user can then train it).