Skip to content

R scripts for datamining. Goal is to create R scripts that combine good datamining methods and knitr to generate datamining result reports based on the CRISP datamining methodology

Notifications You must be signed in to change notification settings

hugokoopmans/dataMineR

Repository files navigation

dataMineR

Update 7/6/2013

In step 1, we now render tables in extended (Pandoc) markdown. This means that Pandoc can be used to output either pdf or html in a well-formatted way.

Update

We are now transforming from latex to markdown

R scripts for datamining.

This project aims to efficiently implement the CRISP datamining cycle. Initially we focus on supervised models for dichotomous class predictive models.

The project tries to combine R scripts with knitr in R-studio in such a way that we walk through the CRISP phases and deliver a scoring model in the most efficient way. In the meantime we would like to document our steps in a report that is nice enough to be used a reference of the work been done.

From wikipedia: CRISP-DM breaks the process of data mining into six major phases:

  • Business Understanding
  • Data Understanding
  • Data Preparation
  • Modelling
  • Evaluation
  • Deployment

In this project we focus on automating the steps of Data Understanding, Data Preparation, Modelling, Evaluation.

HowTo

Put your data file into the data directory. Point the data-analysis.R to the right file name, to do this chage line

path2file <- "../data/ano_churn_data.Rdata"

to point to your filename.

Be sure to have your target variable named "target" or adjust target_name in file Data-Prperation-Report.Rmd to whatever you want to predict in your dataset

target defenition

target_name <- 'target'

Run step by step

About

R scripts for datamining. Goal is to create R scripts that combine good datamining methods and knitr to generate datamining result reports based on the CRISP datamining methodology

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published