Merge pull request #2 from DavidChouinard/patch-1

Fixed typo with verb "doesn't"

HW4.ipynb

@@ -750,7 +750,7 @@
             "\n", 
             "Recall the notions of regularization introduced in class. We want to reduce the variance in our estimates, so we pull our estimates in toward a conservative point in a way that strongly corrals in estimates when there is very little data, but allows the data to speak when there is a lot. This can be shown as equivalent to adding in a `reg` amount of bayesian prior, as Joe has alluded to in class. \n", 
             "\n", 
-            "A good value of the regularizer is intuitively one that dosent affect the similarity when the common support is high ~ 10, but has a large effect when the support is small. In this case, values of 2-4 are good. Usually, the value of `reg` is determined using cross-validation, but for the sake of simplicity we will generally set it to 3.\n", 
+            "A good value of the regularizer is intuitively one that doesn't affect the similarity when the common support is high ~10, but has a large effect when the support is small. In this case, values of 2-4 are good. Usually, the value of `reg` is determined using cross-validation, but for the sake of simplicity we will generally set it to 3.\n", 
             "\n", 
             "We define a function `shrunk_sim` which takes the `sim` and `n_common` obtained from the database, and shrinks the similarity down using the regularizer `reg`."
           ], 
@@ -2310,4 +2310,4 @@
       "metadata": {}
     }
   ]
-}
+}