In-class exercise

1
In-class exercise

• Use the Job dataset
• Caution clean the spreadsheets first before
  using TANAGRA
• Make sure outcome value is discrete (it already
  is for the Job dataset)
• Make sure there are no empty columns
• Fix Experience column (convert 0 values into a
  discrete description)

2
In-class exercise

• Run raw, categorized, and standard data against
  all algorithms we have tried so far
• Note neural nets work on continuous inputs
  only, naïve bayes works on discrete inputs only
• For the raw data, have several runs trying only
  continuous data, only discrete data, or both
• Try three different decision tree algorithms CRT,
  ID3, C4.5
• Conclude which methods work well for which
  datasets