ÃÛ¶¹ÊÓƵ

Transcript

Hello, in this video we are going to show you how to ingest the CourseLuma demo dataset into ÃÛ¶¹ÊÓƵ Experience Platform. We will be using this data for the remainder of the course lessons to help showcase Data Science Workspace. If you are following along, let’s start by making sure we are in the sandbox we want to ingest the data in. In my case, this is the development sandbox called Data Science Workspace Course. In order to ingest the demo data, we need to start by creating a schema to define the data structure. Navigate to the Data Management section in Experience Platform and select Schemas. Next, we want to select the Create Schema button, then select XDM Experience Event option. The data we are using is Luma Web Data, and we want this schema to capture the time-stamped behaviors of our web users, so the Experience Event option makes the most sense here. After naming our schema and providing a description, we need to add some schema field groups, previously known as mixins, to capture our data. Let’s start by adding the Consumer Experience Event field group. Once the Consumer Experience Event field group has been added, you will see a number of additional fields highlighted in the canvas. These are the fields that were added to our schema. Our web data also has some IDs associated with it, so we want to add an additional field group to capture our end user IDs. Let’s add the End User ID Details field group as well. Now that we have a schema to capture our data, we need to set a primary identity. To do this, let’s select our End User ID Details field group and navigate our schema to find an MCID column. Once we open the MCID column, select the ID field from within the column. The Field Properties sidebar opens. Scroll down and select the Identity checkbox, then select the Primary Identity checkbox, followed by selecting the ECID namespace from the Identity Namespace dropdown. Then select Apply, followed by Saving your changes. There you have it. We now have our schema and are ready to create a new dataset. To create a new dataset, select Datasets from within the Data Management tab. Then in the top right corner, select Create Dataset. Two options are provided. We want to select the Create Dataset from Schema option. Once you select Create Dataset from Schema, a list of schemas is provided. Select the schema we just made, then select Next. You will be asked to name the dataset and provide a description. Once complete, we are redirected to the Dataset Activity tab for our dataset. If you have not already, download the LumaDemo dataset from this module. In this demo, we are going to ingest our data by dragging and dropping the demo files into the Add Data section. Upon doing so, a batch ID is created. You’ll need to wait a couple minutes for the batch to finish processing. Going back to our architecture diagram, we can see that we used batch mode and our data is going to be mapped to our XDM schema, then added to the data lake where we can query and transform the data using Data Science Workspace. Once complete, a successful status is displayed. Before we finish, let’s quickly preview the data by selecting the Preview button from the Dataset Activity tab. We can see that our data was added to our schema and take a quick look at the course data that’s going to be used. With that, we have our course data in the data lake and we are ready to proceed with the remainder of the course. Thanks for watching!

Transcript

Hello, in this video we are going to show you how to create the course training, scoring, and scoring results datasets in ÃÛ¶¹ÊÓƵ Experience Platform. We will be using this data to train and score a model in a subsequent tutorial. We will then store the model results in a scoring results dataset that is enabled for real-time customer profile. If you are following along, let’s start by navigating to the dataset tab and select Create New Dataset. We want to create our dataset from a schema. The training and scoring datasets use the same schema as our Luma web data. All we have to do is select the schema we made in the course data ingestion tutorial and provide a name and description for the new dataset. I’ll name mine Data Science Course Training Dataset, followed by doing the same steps to create a Data Science Course Scoring dataset. Next, we want to select the notebooks tab and open up JupyterLab. We have provided a notebook named Data Science Training and Scoring Ingestion. Import and open the provided Spark notebook. This notebook is going to split our Luma web data into a training and scoring dataset and then write the split data to the two datasets we just made. In order for the notebook to run, you need to provide the dataset IDs for your training, and Luma dataset. Once you have run all the cells, check to make sure your training and scoring dataset contain the new data. The training dataset should be considerably larger than the scoring dataset. Next we want to create a new schema for our scoring results dataset. We will use this schema and dataset to store the model output. Select the Create Schema button, then select XDM Individual Profile as the option. This schema will need a custom field group, so we want to select the Create New Field Group option. Provide a display name and description for the new field group, then select Add Field Groups. After adding the field group, select the Add icon to create a new field and open the Field Properties menu. Our propensity model is going to output two values that I want to capture, a primary identifier and a boolean value for will order. Let’s start by creating the boolean value for will order. The field name should use camelcase. The display name is used in the UI, such as the Segment Builder. We want the type to be set to boolean and the default value for our boolean to be set to false. Select Apply and the new field is created, but we are not done yet. Next we need to capture our customer identifier. Select the Add option and create the Experience Cloud ID field, or ECID for short. Set the type to String, then scroll down and check the required box. We must have this value in order to create a profile and attach our will order prediction value. Because this is our identifier, we also need to select the Identity checkbox, followed by selecting the Primary Identity checkbox, and then select the ECID namespace from the Identity Namespaces dropdown. After finishing, select Apply. Now that we have our custom field group to capture the model output data, we can name the schema and provide a description. But before we save the schema, we want to enable it for profile by selecting the Profile toggle. In order for data to be ingested into profile, both the schema and dataset must be enabled for profile. Once complete, select Save to finish the schema. We are now ready to create the Profile Prediction dataset. Following the same steps we did to create the Training and Scoring dataset, we want to create the Scoring Results dataset with our new schema. I’ll name mine Data Science Course Model Prediction dataset. Once the dataset has been created, select the Profile toggle. This dataset will store all our model output scoring results, and because the schema and dataset are enabled for profile, the results will also be sent to real-time customer profile. With that, we have the required datasets to create the Course Propensity model. Thanks for watching.