The forecasting of product and service demand, coupled with the extraction of behavioural insights, facilitates greater efficiencies in providing services to the end customer. This case study explores the development of a demand forecasting and behavioural insights model for one of our clients in the digital media vertical.
The client is an online digital media platform that helps creators and brands produce, host, share, and monetise content. They required the forecasting of future media consumption demand in order to optimise release and advertisement timings. Furthermore, behavioural insights were warranted in the aim of developing a deeper understanding of consumption patterns and maximising the amount of media consumption on a per category and release interval basis.
Our approach involved the development of a demand forecasting and behavioural insights model that enabled the optimisation and prediction of media demand. Firstly, an extensive magnitude of media and consumption data were amalgamated and batch processed on AWS EC2 instances, formulating an actionable data structure and identifying the key consumption behavioural patterns for each media category and release interval. A demand forecasting model was subsequently developed, trained, and deployed on the AWS SageMaker platform. Furthermore, the model understanding was enhanced through our expertise in interpretable machine learning methods – driving insights from typically black-box predictive models.
The machine learning and analytics solution enabled the client to optimise and forecast media consumption demand based on product categories and release timings. Moreover, our analysis facilitated a greater understanding of consumption patterns for more effective advertisement timings.