Judah Phillips on Machine Learning in Marketing
By: Shreya Jain
“Machine intelligence is the last invention that humanity will ever need to make.” – Nick Bostrom
The marketing big data ecosystem can be greatly leveraged upon by the use of machine learning. It definitely aids businesses in making better decisions and any firm employing it well will have a competitive advantage in the long run. The increased level of automation will make businesses smarter, savvier and well-informed!
Judah Phillips is an award-winning entrepreneur, author and a leader in the field of data science for 20 years. His best-sellers namely Ecommerce Analytics, Building a Digital Analytics Organization, and Digital Analytics Primer, exemplify the power of data science and analytics. He is the CEO and co-founder of Squark, a firm that focuses on providing deployable machine learning predictions to companies aiming at improving marketing effectiveness
Marketing Analytics is largely viewed as a descriptive and diagnostic tool. However, Judah Phillips emphasizes on the need to extend it as a predictive and prescriptive tool, and adopt a proactive approach towards it. This allows companies to know the full story behind the data, starting from “What happened in the past and why?” to “What is likely to happen in future”.
Phillips defines supervised machine learning as the process of building models based on well labelled past data and applying that learning to predict future outcomes. It assumes that the future is similar to the past. In marketing, it is extensively used for churn rate prediction, promotional optimization and customer segmentation. This helps companies devise better marketing campaigns, target the right audiences, reduce risk and improve return on investments. Binary and multinomial classification, regression model and times series forecasting are a few important predictive models. The model that fits the data most appropriately is selected on the basis of the number of entities to be modelled, the number of periods and seasonality in the data.
Phillips talks about how Squark is perfect for users who want to capitalize on data capabilities but aren’t experts at coding. Building models and gaining insights on Squark only takes a few clicks, and no code at all! The process begins with creating a project, loading the required datasets, selection of dependent and independent variables, generating the model and interpreting it. It employs feature engineering wherein powerful machine learning algorithms are used to structure the predictive modeling problems. It validates the accuracy of the model based on the associated error gradients like deviance, class error, logarithmic loss, mean standard error.
It also provides the statistical significance of the different predictors of the dependent variables. Additionally, the confusion matrix provides the accuracy of the model. The advantages of Squark are easy navigability, zero coding requirements and it is time & cost effective.
Businesses that employ machine learning will be able to more prepared for changes, have better response systems in place and this new technology will make them more productive and efficient!
Let us know what you think the future of machine learning holds and how it plays an important role in your field.