Minor changes to Lab 9

Author: maalvarezl

Lab 9 - Scalable Generalized Linear Models.ipynb

@@ -210,7 +210,7 @@
    "source": [
     "### Question 1\n",
     "\n",
-    "The variables season, yr, mnth, hr, holiday, weekday and weathersit are categorical variables that have been treated as continuous variables. In general this is not optimal since we are indirectly imposing a geometry or order over a variable that does not need to have such geometry. For example, the variable season takes values 1 (spring), 2 (summer), 3 (fall) and 4 (winter). Indirectly, we are saying that the distance between spring and winter (1 and 4) is larger than the distance between spring (1) and summer (3). There is not really a reason for this. To avoid this imposed geometries over variables that do not follow one, the usual approach is to transform categorical features to a representation of one-hot encoding. Use the [OneHotEncoderEstimator](http://spark.apache.org/docs/2.3.2/api/python/pyspark.ml.html#pyspark.ml.feature.OneHotEncoderEstimator) over the Bike Sharing Dataset to represent the categorical variables. Using the same training and test data compute the RMSE over the test data using the same Poisson model. \n",
+    "The variables season, yr, mnth, hr, holiday, weekday, workingday and weathersit are categorical variables that have been treated as continuous variables. In general this is not optimal since we are indirectly imposing a geometry or order over a variable that does not need to have such geometry. For example, the variable season takes values 1 (spring), 2 (summer), 3 (fall) and 4 (winter). Indirectly, we are saying that the distance between spring and winter (1 and 4) is larger than the distance between spring (1) and summer (3). There is not really a reason for this. To avoid this imposed geometries over variables that do not follow one, the usual approach is to transform categorical features to a representation of one-hot encoding. Use the [OneHotEncoderEstimator](http://spark.apache.org/docs/2.3.2/api/python/pyspark.ml.html#pyspark.ml.feature.OneHotEncoderEstimator) over the Bike Sharing Dataset to represent the categorical variables. Using the same training and test data compute the RMSE over the test data using the same Poisson model. \n",
     "\n",
     "### Question 2\n",
     "\n",
@@ -244,7 +244,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.7.1"
   }
  },
  "nbformat": 4,