naive bayes classification matrix

I created a Naive Bayes mining structure.  SSAS produced the following classification matrix

Predictedpeak (Actual)not (Actual)

peak142 681

not311 8675

I wanted to research the records that were incorrectly classified.  So I created a simple DMX query (which happened to be the first DMX query I have ever written) to show those records.  I joined the mining model to the same data source view that the mining model was created from while joining on its primary key.  Even though the classification matrix indicates that there were records accurately predicting the target of 'peak', the query below is showing every record (67801 total records) as predicting the other value 'not'.  I am a newbie in data mining, so I am confident that I am doing something stupid.

SELECT

  [Bottom Peak 1].[Wave Peak],
  t.[m3dtWaveStart],
  t.[wavePeak]
From
  [Bottom Peak 1]
PREDICTION JOIN
  OPENQUERY([Forex],
    'SELECT
      [m3dtWaveStart],
      [wavePeak]
    FROM
      [dbo].[H1M3wavePeaks]
    ') AS t
ON
  [Bottom Peak 1].[m3dt Wave Start] = t.[m3dtWaveStart]