v0.2.11:

Version 0.2.11:

New Features

• added (very limited) sklearn.Pipeline support. You can pass a Pipeline as
  model parameter as long as the pipeline either:
  1. Does not add, remove or reorders any input columns
  2. has a .get_feature_names() method that returns the new column names
     (this is currently beings debated in sklearn SLEP007)
• added cutoff slider to CumulativePrecisionComponent
• For RegressionExplainer added ActualVsColComponent and PredsVsColComponent
  in order to investigate partial correlations between y/preds and
  various features.
• added index_name parameter: name of the index column (defaults to X.index.name
  or idxs.name). So when you pass index_name="Passenger", you get
  a "Random Passenger" button on the index selector instead of "Random Index",
  etc.

Bug Fixes

• Fixed a number of bugs for when no labels are passed (y=None):
  • fixing explainer.random_index() for when y is missing
  • Hiding label/y/residuals selector in RandomIndexSelectors
  • Hiding y/residuals in prediction summary
  • Hiding model_summary tab
  • Removing permutation importances from dashboard

Improvements

• Seperated labels for "observed" and "average prediction" better in tree plot
• Renamed "actual" to "observed" in prediction summary
• added unique column check for whatif-component with clearer error message
• model metrics now formatted in a nice table
• removed most of the loading spinners as most graphs are not long loads anyway.

v0.2.10: cats parameter now take arbitrary list of onehot cols

Version 0.2.10:

New Features

• Explainer parameter cats now takes dicts as well as a list where you can specify
  your own groups of onehotencoded columns.
  - e.g. instead of passing cats=['Sex'] to group ['Sex_female', 'Sex_male', 'Sex_nan']
  you can now do this explicitly: cats={'Gender'=['Sex_female', 'Sex_male', 'Sex_nan']}
  - Or combine the two:
  cats=[{'Gender'=['Sex_female', 'Sex_male', 'Sex_nan']}, 'Deck', 'Embarked']

v.0.2.9.1: forgot version bump

v0.2.9.1

bumped version

v.0.2.9: reducing custom boilerplate code

Version 0.2.9:

New Features

• Reducing boilerplate: You don't have to pass the list of subcomponents in self.register_components()
  anymore: it will infer them automatically from self.__dict__.

Improvements

• Reducing boilerplate: ExplainerComponents now automatically stores all parameters to attributes
• ExplainerComponents now automatically stores all parameters to a ._stored_params dict
• ExplainerDashboard.to_yaml() now support instantiated tabs and stores parameters to yaml
• ExplainerDashboard.to_yaml() now stores the import requirements of subcomponents
• ExplainerDashboard.from_config() now instantiates tabs with stored parameters
• ExplainerDashboard.from_config() now imports classes of subcomponents

Other Changes

• added docstrings to explainer_plots
• added screenshots of ExplainerComponents to docs
• added more gifs to the documentation

v0.2.8.1: shap dependency >=0.36

bump version

v0.2.8: load dashboards from config

Version 0.2.8:

Breaking Changes

• split explainerdashboard.yaml into a explainer.yaml and dashboard.yaml

  • generate with explainer.to_yaml("explainer.yaml") and db.to_yaml("dashboard.yaml")

• Changed UI of the explainerdashboard CLI to reflect this

• This change makes it possible to have automatic rebuilds and redeploys
  when an modelfile, datafile or configuration file changes. See deployment documentation.

New Features

• Load an ExplainerDashboard from a configuration file with the classmethod .from_config()
  e.g. : ExplainerDashboard.from_config("dashboard.yaml")

0.2.7 small fix

small bug fix in dashboard.to_yaml

new CLI tool

Version 0.2.7:

New Features

• explainer.dump() to store explainer, explainer.from_file() to load
  explainer from file
• Explainer.to_yaml() and ExplainerDashboard.to_yaml() can store the
  configuration of your explainer/dashboard to file.
• explainerdashboard CLI:
  • Start an explainerdashboard from the command-line!
  • start default dashboard from stored explainer : $ explainerdashboard run explainer.joblib
  • start full configured dashboard from config: $ explainerdashboard run explainerdashboard.yaml
  • build explainer based on input files defined in .yaml
    (model.pkl, data.csv, etc): $ explainerdashboard build explainerdashboard.yaml
  • includes new ascii logo :)

Improvements

• If idxs is not passed use X.index instead
• explainer.idxs performance enhancements
• added whatif component and tab to InlineExplainer
• added cumulative precision component to InlineExplainer

v0.2.6: simpler imports

Improvements

• more straightforward imports: from explainerdashboard import ClassifierExplainer, RegressionExplainer, ExplainerDashboard, InlineExplainer
• all custom imports (such as ExplainerComponents, Composites, Tabs, etc)
  combined under explainerdashboard.custom:
  from explainerdashboard.custom import *

v0.2.5

version 0.2.5:

New Features

• New dashboard tab: WhatIfComponent/WhatIfComposite/WhatIfTab: allows you
  to explore whatif scenario's by editing multiple features and observing
  shap contributions and pdp plots. Switch off with ExplainerDashboard
  parameter whatif=False.
• New login functionality: you can restrict access to your dashboard by passing
  a list of [login, password] pairs:
  ExplainerDashboard(explainer, logins=[['login1', 'password1'], ['login2', 'password2']]).run()
• Added 'target' parameter to explainer, to make more descriptive plots.
  e.g. by setting target='Fare', will show 'Predicted Fare' instead of
  simply 'Prediction' in various plots.
• in detailed shap/interaction summary plots, can now click on single
  shap value for a particular feature, and have that index highlighted
  for all features.
• autodetecting Google colab environment and setting mode='external'
  (and suggesting so for jupyter notebook environments)
• confusion matrix now showing both percentage and counts
• Added classifier model performance summary component
• Added cumulative precision component

Improvements

• added documentation on how to deploy to heroku
• Cleaned up modebars for figures
• ClassifierExplainer asserts predict_proba attribute of model
• with model_output='logodds' still display probability in prediction summary

Other Changes

• removed monkeypatching shap_explainer note