Difference between revisions of "Banjo"

From wiki
Jump to: navigation, search
 
(One intermediate revision by the same user not shown)
Line 6: Line 6:
 
  module load banjo
 
  module load banjo
  
= Quirks of usage =
+
= Usage =
  
It does not handle NULL values very well. In fact, it may even ignore them.
+
* A canonical input dataset would be one with variables as columns and observations as rows.
 
+
* A settings file must contain all this information and other parameters to way the program is run.
Certainly, if there are null values, all rows must have the same number.
+
* A structure file is also required, which gives a graph structure.
 +
* There is no data imputation on NULL values: either the columns and rows which contain them are deleted, or they are given an encoding of their own.
  
 
== Files which must be in input folder ==
 
== Files which must be in input folder ==

Latest revision as of 10:47, 6 October 2017

Introduction

Banjo is Bayesian Network software with a very range of applicability in terms of data sets.

It is coded in java and can be activiated on the cluster with

module load banjo

Usage

  • A canonical input dataset would be one with variables as columns and observations as rows.
  • A settings file must contain all this information and other parameters to way the program is run.
  • A structure file is also required, which gives a graph structure.
  • There is no data imputation on NULL values: either the columns and rows which contain them are deleted, or they are given an encoding of their own.

Files which must be in input folder

This error may come up

(EdgesAsArrayWithCachedStatistics.loadFromFile) Cannot find the structure file: 'static.mandatory.str' in directory '/storage/home/users/njb8/ex99/input'.

This refers to a structure file and demands more in depth knowledge of Banjo.