diff --git a/learning_curve.py b/learning_curve.py
index 2364f2c..3f7d028 100644
--- a/learning_curve.py
+++ b/learning_curve.py
@@ -8,17 +8,30 @@
 
 data = load_digits()
 print data.DESCR
-num_trials = 10
-train_percentages = range(5,95,5)
+num_trials = 100
+train_percentages = range(5, 95, 5)
 test_accuracies = numpy.zeros(len(train_percentages))
 
-# train a model with training percentages between 5 and 90 (see train_percentages) and evaluate
-# the resultant accuracy.
-# You should repeat each training percentage num_trials times to smooth out variability
-# for consistency with the previous example use model = LogisticRegression(C=10**-10) for your learner
+# train a model with training percentages between 5 and 90 (see
+# train_percentages) and evaluate the resultant accuracy.
+# You should repeat each training percentage num_trials times to smooth out
+# variability for consistency with the previous example use
+# model = LogisticRegression(C=10**-10) for your learner
 
-# TODO: your code here
+# create a model
+model = LogisticRegression(C=10**-100)
 
+# for loop so the things happen for the correct values enough times
+for n in range(len(train_percentages)):
+    for i in range(num_trials):
+        # Split the data and train on some of it, store data correctly
+        X_train, X_test, y_train, y_test = train_test_split(data.data,
+                                                            data.target,
+                                                            train_size=train_percentages[n])
+        model.fit(X_train, y_train)
+        test_accuracies[n] = test_accuracies[n] + model.score(X_test, y_test)
+
+# create  plot and put the correct things on it
 fig = plt.figure()
 plt.plot(train_percentages, test_accuracies)
 plt.xlabel('Percentage of Data Used for Training')
diff --git a/questions.txt b/questions.txt
new file mode 100644
index 0000000..575f0a9
--- /dev/null
+++ b/questions.txt
@@ -0,0 +1,7 @@
+1. The general trend in the curve is positive.
+
+2. At only 10 trials, almost the entire curve seems to be noisy. Typically, accuracy from training on 5% on the test is lower than on 90%, but everything between is pretty inconsistent.
+
+3. 100 trials makes a fairly smooth curve, but it's still pretty piecewise.
+
+4. The scale of accuracy on the test set gets smaller with a smaller C value and the entire curve becomes less smooth.