Page tree
Skip to end of metadata
Go to start of metadata

The Interaction Studio Data Science Workbench gives data scientists an easy and powerful way to access the rich customer data stored in Interaction Studio. Interaction Studio maintains and updates this data automatically, so it is always the most up-to-date information about your customer and visitors. Your data scientists will be able to work with familiar tools that Interaction Studio automatically manages and updates. These tools incorporate modern machine learning so your data scientists can analyze your customer data, leverage Interaction Studio notebooks and models, then use the model results for Interaction Studio segments and machine learning algorithms.


This Article Explains

This article gives you an overview of how Data Science Workbench can help you access and work with Interaction Studio data. 

Sections in this Article

Access and Work with Interaction Studio's Rich Customer Data

Interaction Studio’s Data Science Workbench gives your data scientists a web-based approach to access, analyze and enhance the rich customer data collected by Interaction Studio on your customers and visitors. For every visitor, customer, and account, Interaction Studio maintains a unified profile which can contain a deep contextual understanding of that person’s website and mobile app behavior, email interaction, campaign interaction, survey responses, and purchases, along with attribute data collected through any feed or system Interaction Studio integrates with.

In addition, Interaction Studio’s robust Identity System unites disparate customer information into that unified profile. The result is that Interaction Studio’s behavioral data is much richer than any other single system, since it combines contextual understanding of your business with time-spent and true engagement data.

NOTE

A maximum of two users can be enabled on the Data Science Workbench.

Interaction Studio Maintains and Updates Your Data

So much of a data scientist’s work is in data preparation. Interaction Studio stores data in a way that is optimized for performance, so it can be difficult to use this data for ad-hoc analysis. Rather than forcing data scientists to waste time translating the native format, Interaction Studio maintains a copy of your data in a convenient relational database. This data warehouse is updated daily and provides an easy and efficient way for data scientists to quickly get at the data they need.

The Data Science Workbench comes pre-loaded with a library with convenient functions which allow your data scientists to load data from the Interaction Studio Data Warehouse into a Spark DataFrame structure that is accessible in R. Pre-canned queries transform the data into comfortable, well-labeled, and documented Spark DataFrame structures that can be interacted with in RStudio as easily as native R DataFrame objects. These queries include requests for user information and history, product catalogs, and campaign metrics along with raw A/B test results.


Work with Familiar Tools, Maintained and Updated for You   

Data Science Workbench users are provided with a dedicated high-performance computing server instance to access Interaction Studio’s data through a safe and secure data warehouse. The cluster is pre-installed with a suite of tools (familiar to your data scientists) which include Apache Spark, Rstudio Server, and H2O. Interaction Studio automatically manages, maintains, and updates the cluster and software versions.


Analyze Data with Modern Machine Learning

Your data scientists can run data transformations, numerical simulations, statistical modeling, and data visualizations on source data by writing R code, which is the same way Interaction Studio’s own data scientists do this same work.


Leverage Interaction Studio Statistics, Pre-built Models, and Notebooks

Interaction Studio provides many out-of-the-box options for data scientists. In addition to supervised and unsupervised algorithms available through H20, which is accessible programmatically in RStudio or with a GUI, H2O provides many models including deep learning, generalized linear models, and isolation forest. Data scientists also have access to additional code to perform statistical analysis with Interaction Studio’s Bayesian Stats package.

Leverage Model Results in Segments and Machine Learning

After data scientists create a model in the Data Science Workbench, scores can be uploaded back into Interaction Studio through the ETL system as custom user attributes. For instance, based on your analysis you could flag particular customers as “high value,” “seasonal buyers,” or “likely to churn.” An attribute is part of the Interaction Studio data model, so these predictive value attributes can be used to segment and target customers. Additionally, attributes can be applied as custom features in Interaction Studio’s Einstein Decisions algorithm.