amlr 0.3.6

amlr - Auto Machine Learning Report

AMLR - Auto Machine Learning Report

Create a bealtifull Machine Learning Report with One-Line-Code

---

Main Features:

• Exploratory Data Analisys
  • Dataset Configuration
    • Shape
    • Detect number of classes (Bernoulli or binary for while)
    • Automatically Duplicate Observations dropped
    • You can drop Duplicate Observations manually as well
    • Exclude automatically features with highest frequencies (Names, IDs, FW keys etc)
  • Regression Analysis
  • Automatic Balance Classes
  • Correlation Analysis
  • Detecting Multicollinearity with VIF
  • Residual Analisys
• Grid - Hyperparameter optimization
• Partial dependence plot (PDP)
• Individual Conditional Expectation (ICE)
• Variable Importance by Model
• AML - Partial Dependence
• Ensemble - (ICE) Individual Condition Expectation
• Correlation Heatmap by Model
• Model Performance
  • Analytical Performance Modeling
  • Comparative Metrics Table with:
    • Overall ACC
    • Kappa
    • Overall
    • RACC
    • SOA6(Landis & Koch)
    • SOA6(Fleiss)
    • SOA6(Altman)
    • SOA6(Cicchetti)
    • SOA6(Cramer)
    • SOA6(Matthews)
    • TNR Macro
    • TPR Macro
    • FPR Macro
    • FNR Macro
    • PPV Macro
    • ACC Macro
    • F6 Macro
    • TNR Micro
    • FPR Micro
    • TPR Micro
    • FNR Micro
    • PPV Micro
    • F6 Micro
    • Scott PI
    • Gwet AC6
    • Bennett S
    • Kappa Standard Error
    • Kappa 96% CI
    • Chi-Squared
    • Phi-Squared
    • Cramer V
    • Chi-Squared DF
    • 96% CI
    • Standard Error
    • Response Entropy
    • Reference Entropy
    • Cross Entropy
    • Joint Entropy
    • Conditional Entropy
    • KL Divergence
    • Lambda B
    • Lambda A
    • Kappa Unbiased
    • Overall RACCU
    • Kappa No Prevalence
    • Mutual Information
    • Overall J
    • Hamming Loss
    • Zero-one Loss
    • NIR
    • P-Value
    • Overall CEN
    • Overall MCEN
    • Overall MCC
    • RR
    • CBA
    • AUNU
    • AUNP
    • RCI
    • Pearson C
    • CSI
    • ARI
    • Bangdiwala B
    • Krippendorff
    • Alpha
  • The Best Algorithms Table
    • Automatically chooses the best model based on the metrics above
  • Confusion Matrix for all Models
  • Feature Importance for all models
  • Save all Models into a Pickle file

---

How to Install

sudo apt-get install default-jre
pip install amlr

---

How to use

sintax:

from amlr import amlr as rp
import webbrowser

rp = rp.report()
rp.create_report(dataset='data/titanic-passengers.csv', target='Survived')

webbrowser.open('report/index.html')

We tested with the following Data Sets

• Classic dataset on Titanic disaster
  • Bernoulli Distribution Target or Binary Classification
  • Download here: Titanic

Output

See the output here

This is an example of the test made with the Titanic Dataset;

---

enjoi!