Show HN: Automatic Markdown Reports from Machine Learning Experiments

Show HN: Automatic Markdown Reports from Machine Learning Experiments

Build Status
PyPI version
PyPI pyversions

Automated Machine Discovering out

mljar-supervised is an Automated Machine Discovering out python tools. It will recount ML objects for:

  • binary classification,
  • multi-class classification,
  • regression.

What’s applicable in it?

  • mljar-supervised creates markdown evaluations from AutoML coaching. The instance of AutoML leaderboard abstract:

AutoML leaderboard

The instance for Decision Tree abstract:
Decision Tree summary

The instance for LightGBM abstract:
Decision Tree summary

  • This tools is computing Baseline to your information. So you’re going to know in the event you’d love Machine Discovering out or not! You are going to know the scheme applicable are your ML objects evaluating to the Baseline. The Baseline is computed in line with prior class distribution for classification, and easy imply for regression.
  • This tools is coaching simple Decision Timber with max_depth <= 5, in order that that you can maybe presumably with out considerations visualize them with good dtreeviz to higher understand your information.
  • The mljar-supervised is the utilization of simple linear regression and include its coefficients throughout the abstract file, in order that that you can maybe presumably affirm which points are aged primarily essentially the most throughout the linear mannequin.
  • It's the utilization of an favorable blueprint of algorithms: Random Forest, Further Timber, LightGBM, Xgboost, CatBoost (Neural Networks would maybe be added quickly).
  • It will create points preprocessing, love: missing values imputation and altering categoricals. What's further, it can furthermore deal with goal values preprocessing (Which that you can not think about how progressively it is needed!). As an illustration, altering convey goal into numeric.
  • It will tune hyper-parameters with no longer-so-random-search algorithm (random-search over outlined blueprint of values) and hill climbing to graceful-tune remaining objects.
  • It will compute Ensemble in line with greedy algorithm from Caruana paper.
  • It cares about explainability of objects: for each algorithm, the function significance is computed in line with permutation. Furthermore, for each algorithm the SHAP explanations are computed: function significance, dependence plots, and choice plots (explanations would maybe be switched off with explain_level parameter).

Prompt instance

There might be a simple interface out there with match and predict concepts.

import pandas as pd
from supervised.automl import AutoML

df = pd.read_csv("", skipinitialspace=Moral)

X = df[df.columns[:-1]]
y = df["income"]

automl = AutoML(results_path="directory_with_reports")
automl.match(X, y)

predictions = automl.predict(X)

For particulars please affirm AutoML API Doctors.


Set up

From PyPi repository:

pip arrange mljar-supervised

From supply code:

git clone
cd mljar-supervised
python arrange

Set up for mannequin

git clone
virtualenv venv --python=python3.6
supply venv/bin/activate
pip arrange -r necessities.txt
pip arrange -r requirements_dev.txt


The mljar-supervised is an beginning up-source mission created by MLJAR. We care about ease of use throughout the Machine Discovering out.
The presents an exceptional and easy person interface for establishing machine finding out objects.


Please enter your comment!
Please enter your name here