I would recommend starting with a spreadsheet-sized dataset (no more than a few thousand records) where you want to predict one of the columns, use binary decis

I would recommend starting with a spreadsheet-sized dataset (no more than a few thousand records) where you want to predict one of the columns, use binary decision trees to try to predict it's value. Use either Azure ML Studio, or Jupyter with the Sci-Kit Learn library, depending on your comfort level with programming.