Implementing MLOps in the Enterprise

Chapter 9. Solutions for Advanced Data Types

This chapter delves into the intricacies of data analysis and interpretation, focusing on modern techniques and approaches in time series analysis, NLP, video, and image classification. It aims to comprehensively discuss advanced data types and their applications in tackling complex problems for seasoned data scientists as well as beginners.

The chapter discusses the challenges and options associated with data processing and model selection, particularly concerning time series data. We’ll explore different types of solutions, weigh the trade-offs, and discuss specific considerations in the field of MLOps. We will broaden our scope to include various platforms such as AWS, GCP, Hugging Face, and Apple’s CreateML. Each platform offers a unique set of tools and services that can effectively cater to different needs and preferences. By providing an unbiased comparison of these platforms and discussing their pros and cons, we aim to help you make a well-informed decision.

To get a sense of the variety of data types MLOps developers use, look at Apple’s CreateML interface in Figure 9-1. As illustrated, there are categories in Image, Video, Motion, Sound, Text, and Tables.

We’ll return to CreateML toward the end of the chapter.

Figure 9-1. Apple’s CreateML Project Types

ML Problem Framing with Time Series

An inflection point critical decision ...