DriveML

Background

DriveML package for automated machine learning especially in the classification context. DriveML saves a lot of effort required for data preparation, feature engineering, model selection and writing lengthy codes in a programming environment such as R. Overall, the main benefits of DriveML are in development time savings, reduce developers errors, optimal tuning of machine learning models and reproducibility.

DriveML Framework:

DriveML is a series of functions such as AutoDataPrep, AutoMAR, autoMLmodel. DriveML automates some of the most difficult machine learning functions such as data cleaning, data transformations, feature engineering, model training, model validation, model tuning and model selection.

Functionalities of DriveML

Three key features of DriveML : Pre-processing, ML Techniques and Model interpretations

1. Automatic data preparation
   • AutoDataPrep function to generate a novel features based on the functional understanding of the dataset
2. ML Model Optimization
   • autoMLmodel function to develop baseline machine learning models using regression and tree based classification techniques
3. Model interpretations
   • autoMLReport function to print the machine learning model outcome in HTML format

Installation

Install from CRAN within R using:

install.packages("DriveML")

Install the latest development version of the DriveML from github with:

devtools::install_github("daya6489/DriveML")

Example

Data

In this vignette, we will be using Heart Disease - Classifications data set

Data source UCI

    library("DriveML")
    library("SmartEDA")
    ## Load heart disease dataset 
    data(heart)
    

Overview of the data

Understanding the dimensions of the dataset, variable names, overall missing summary and data types of each variables

## overview of the data; 
    ExpData(data = heart, type = 1)
## structure of the data    
    ExpData(data = heart, type = 2)

Summary of numerical variables

To summarise the numeric variables, you can use following r codes from this pacakge

## Summary statistics by – overall
    ExpNumStat(heart, by = "GA", gp = "target_var", Qnt = seq(0, 1, 0.1), MesofShape = 2, Outlier = TRUE, round = 2)

Graphical representation of all numeric features

## Generate Boxplot by category
ExpNumViz(heart, gp = "target_var", type = 2, nlim = 25, Page = c(2, 2))

## Generate Density plot
ExpNumViz(heart, gp = NULL, type = 3, nlim = 10, Page = c(2, 2))

## Generate Scatter plot
ExpNumViz(heart, target="target_var", nlim = 4, scatter = TRUE, Page=c(2, 1))

Machine learning model using 3 simple steps

Data preparation using autoDataprep function

One function to prepare a input data for machine learning model

# Data Preparation
small_data <- autoDataprep(heart, target = "target_var", missimpute = "default",
                       auto_mar = TRUE, mar_object = NULL, dummyvar = TRUE,
                       char_var_limit = 12, aucv = 0.02, corr = 0.99,
                       outlier_flag = TRUE, interaction_var = TRUE,
                       frequent_var = TRUE, uid = NULL, onlykeep = NULL, drop = NULL)

# Print output on R console
printautoDataprep(small_data)

# Final prepared master data

small_data_t <- small_data$master_data

Machine learning classification model using autoMLmodel function

One function to develop machine learning binary classification model

# DriveML Model development
small_ml_random <- autoMLmodel(small_data_t,  target = "target_var",  testSplit = 0.2,
                      tuneIters = 5,  tuneType = "random",
                      models = "all", varImp = 10,  liftGroup = 10,  maxObs = 10000,  uid = NULL,
                      pdp = T, positive = 1, htmlreport = FALSE, seed = 1991)

# Model summary results
small_ml_random$results

Model comparison results

Test AUC

Variable Importance

Threshold Plot

Present model report using autoMLReport function

Generate a report in html format for the output of autoDataprep and autoMLmodel functions.

autoMLReport(mlobject = small_ml_random, mldata = small_data, op_file = "driveML_ouput_heart_data.html")

Articles

The pre-print version of the paper on DriveML is available at ArXiv at- https://arxiv.org/pdf/2005.00478.pdf.

DriveML paper was presented at the 2nd International Workshop on Data Quality Assessment for Machine Learning at Knowledge Discovery and Data Mining (KDD) conference that was held on14-18 August, 2021, at Singapore. Conference workshop website link- http://data-readiness-kdd-2021.mybluemix.net

References

Boulange, A. (2020) automl: Deep Learning with Metaheuristic. URL:https://CRAN.R-project.org/package=automlr package version 1.3.2

Chen et al. (2020). xgboost: Extreme Gradient Boosting. URL:https://CRAN.R-project.org/package=xgboostr package version 1.0.0.2.

He, X., Zhao, K., & Chu, X. (2020). Automl: A survey of the state-of-the-art. arXiv:1908.00709v4. URL: https://arxiv.org/pdf/1908.00709.pdf

Therneau, T., & Atkinson, B. (2019). rpart: Recursive Partitioning and Regression Trees. URL: https://CRAN.R-project.org/package=rpart r package version 4.1-15.

Wright, M. N., & Ziegler, A. (2017). ranger: A fast implementation of random forests for high dimensional data in C ++ and R. Journal of Statistical Software, 77, 1–17. doi:10.18637/jss.v077.i01

Maher, M., & Sakr, S. (2019). Smartml: A meta learning-based framework for au-tomated selection and hyperparameter tuning for machine learning algorithms. In Advances in Database Technology-EDBT 2019: 22nd International Conference on Extending Database Technology . Lisbon, Portugal.