Topic: Week 8: Building an Artificial Neural Network for Prediction This is the overview of the process that you need to create Drag and drop the dataset and make sure there are no missing values in the attribute columns that you want to make the prediction accordingly. For example, if you have any missing value in the column "age" you can replace it with the average age of all the customers In the "filter example" operator, you can filter the rows that have some conditions. For example, I want to have the rows that their "Churn" column is not missing. See the figure below: "Set role" is used here to determine the value of what attribute you want to predict. In this case, we want to predict whether a customer is churn or loyal. Hence, we consider the "churn" attribute as a "label" Keep in mind that Artificial Neural Network (ANN) only operate on numerical data. But we have other data of data in our dataset. So we need to convert them to numerical. Use the operator "nominal to numerical". Choose the codding type "dummy". You need to split the data because we want to have data for training and testing. The training set is for learning purposes and testing is for validation of what the model has learned. In this case, we assume 70% for training and 30% for testing. ANN operator has many parameters that you can change and see their effect on the performance of the model. The "Applymodel"operatoristobuildthelearnedmodeland receive the testing data for validation. Lastly "performance" operator shows the accuracy of the model You should get the results as figures below: Deth in FiRet Q270 /275 example1) accuracy 85.19s Activity: Download the hotel app dataset from Moodle and apply the neural network to classify if the customer is loyal or churn. Use the cross-validation operator to evaluate the performance of your neural network. Change the parameters of your neural network including the number of hidden layers and decay rate to see the effect on performance. Quiz: Bayes' Rule - Given: P(DW) P(W) - What is P(W dry)