Fixed: Markdown in prediction section of factorization machines to include caveat about result alignment

Author: djarpin

introduction_to_amazon_algorithms/factorization_machines_mnist/factorization_machines_mnist.ipynb

@@ -369,7 +369,9 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "As we can see from the confusion matrix above, we predict 942 images of 0 correctly, while we predict 99 images as 0 when in actuality they aren't, and we miss predicting 38 images of 0 that we should have."
+    "As we can see from the confusion matrix above, we predict 951 images of the digit 0 correctly (confusingly this is class 1).  Meanwhile we predict 165 images as the digit 0 when in actuality they aren't, and we miss predicting 29 images of the digit 0 that we should have.\n",
+    "\n",
+    "*Note: Due to some differences in parameter initialization, your results may differ from those listed above, but should remain reasonably consistent.*"
    ]
   },
   {