BigML finds the top 10 most anomalous instances out of your dataset. They are listed on the left column by order, starting with the instance with the highest anomaly score. In addition, each of the instances shows the anomaly score in orange and the predictors in green, which are ordered by field importance under the orange bar. If you move your mouse over the instances listed on the left side you will see the values on the right side also change, this is because the data in the DATA INSPECTOR column corresponds to the instance selected on the TOP ANOMALIES column.
It is worth mentioning that BigML considers a score of 0.60 as a highly likely anomaly. In the example below, all of them are higher than 0.60 therefore all these instances are considered anomalous.
Anomalies can be analyzed in-depth. Just check the box in the instances you are interested and then click the Create dataset button.
We recommend that you read this blog post to see an example of the Anomaly Detector, and read the Dashboard documentation or the documentation for developers for further details.