Data Science Digest 7

A quick look at three of the major players and some of what they have to offer around machine learning.

Title: Amazon Forecast – Accurate time-series forecasting service, based on the same technology used at Amazon.com, no machine learning experience required

Source: https://aws.amazon.com/forecast/
How: Upload your historical and related data, Amazon machine learning and AI generates various forecasts.
When to use this: When you don’t have the resources, tools or in-house talent to build out a forecasting model and system which can accommodate multiple data series which change over time.
Why it’s helpful: Fully managed service “so no servers to provision or machine learning models to build, train or deploy.” Pay as you go so workable for most budgets.
Suggested application: Product demand planning, financial planning, resource planning.
Business impact or insights to be gained: Leveraging machine learning developed by Amazon, forecasts are more accurate and prepared in much shorter time (e.g., from months to hours).

Title: Google Cloud AutoML (range of solutions)

Source: https://cloud.google.com/automl/
How: Different solutions depending on your need: Sight (AutoML Vision, AutoML Video Intelligence), Language (AutoML Natural Language, AutoML Translation), Structured data (AutoML Tables)
When to use this: Leverage what Google has already developed rather than starting a machine learning project from scratch.
Why it’s helpful: Save time and work from a much more evolved starting point for any new projects or data introduction.
Suggested application: Across images, video, text and data sets, analyze, categorize, identify entities within and/or assess attitudes within your content.
Business impact or insights to be gained: Manage scale and speed to apply machine learning benefits within your organization.

Title: Azure Machine Learning – Enterprise-grade machine learning service to build and deploy models faster

Source: https://azure.microsoft.com/en-us/services/machine-learning/
How: “For open source development at cloud scale with a code-first experience” (Basic) or “Basic + UI + secure and comprehensive machine learning lifecycle management for all skill levels” (Enterprise).
When to use this: When you want your MLOps to integrate with existing DevOps processes.
Why it’s helpful: Support for open-source frameworks and languages such as MLFlow, Kubeflow, ONNX, PyTorch, TensorFlow, Python and R, has both a code-first and a drag-and-drop designer, enhanced security.
Suggested application: Run ML on large data sets, including the energy sector, physics research, financial industry and more.
Business impact or insights to be gained: Model transparency and interpretability with consideration to reduce model bias, compliance and audit trails.

Be sure to subscribe to this blog for more tips. And sign up for our free tools to learn of job opportunities and more.