Driven by machine learning, the explosion of data has many companies feeling like they are being left behind. How can businesses derive value from these new technological developments? This course will address this issue and will help you understand what exactly machine learning and predictive analytics are, what are its limits and its potential risks, and why it may benefit your organization. Using real world case studies and many other examples of current and potential future industry usage, this course will help you better understand why many corporations are adopting, or should be adopting machine learning to better enable their future.
Along the way you will learn the types of problems machine learning can solve, be conversant about the class of algorithms one can use, and the process for creating a successful project that incorporates machine learning.
What am I going to get from this course?
- Provide an in-depth understanding of predictive analytics and machine learning and how they can be used to identify potential new applications
- Identify what's possible with the state of the technology in today's world
- Help you determine how machine learning fits into your organization
- Understand how organizations are using machine learning to conduct their business to sustain competitive advantage
- Understand the limits and risks inherent in applying machine learning and predictive analytics
- Obtain a robust knowledge of the business challenges and strategic rewards of Predictive Analytics and Machine Leaning initiatives
- Be conversant in real world case studies, and the reasons for both their successes and failures
- Understand the process for creating successful Machine Learning and analytics initiatives
Module 1: What is Predictive Analytics?
An Overview of Predictive Analytics
Describes what is predictive analytics and why it is important.
Some Pitfalls to Avoid in Creating Analytics Initiatives
Describes how to successfully incorporate predictive analytics initiatives into your organization. What are the right questions to ask and what are some pitfalls to avoid?
Module 2: What is Machine Learning?
Important Machine Learning Concepts
We define machine learning from several different perspectives and apply one of these perspectives to a simple example. We address what fundamental question does machine learning seek to address and conclude with the differences between machine learning and traditional programming.
Types of Learning Styles
We discuss the different "styles" of machine learning, supervised, unsupervised, and reinforcement, along with numerous examples of their use.
We present a high-level view of the different types of supervised algorithms and an intuitive explanation of how they work.
We present a high-level view of the different types of unsupervised algorithms and an intuitive explanation of how they work.
Other Algorithm Types
We present a high-level view of algorithms which are neither supervised or unsupervised, e.g., reinforcement, or some combination of learning styles, e.g., evolutionary and graphical models, and an intuitive explanation of how they work.
A Visual Algorithm Similarity Map
A visual overview of the different classes of machine learning algorithms
Can Machine Learning be Automated?
We define what is automated machine learning, its rationale, and some current automated learning tools.
Module 3: How does Machine Learning Work and What Can Go Wrong?
How Does Machine Learning Work?
Describes the three main components of machine learning systems and how these systems work
What Can Machine Learning Do and What Can It Not Do?
Describes the type of problems machine learning systems can and cannot currently solve
What Can Go Wrong with Machine Learning?
Discusses the concept of overfitting and underfitting, the bias variance tradeoff, and the implications for the predictive ability of embedded applications.
Module 4: CRISP-DM Process
What is CRISP-DM?
We discuss each element of the CRISP-DM standard for analytics initiatives
Lessons Learned from the Real World
The author discusses his experiences with two multinational companies and the consequences of not carefully adhering to the CRISP-DM process
Some Problems in Using CRISP-DM
The author discusses potential problems that analytic projects may face in their use of the CRISP-DM process
Case Study – Predicting the Future of Coffee in New York City
The author walks you through a real world application, step by step, while discussing the corresponding steps in the CRISP-DM process necessary to make the project a success
Module 5: Predictive Data Analytics Tools
Gartner Data Science Platform Evaluation Guide Criteria
Describes the set of criteria used by Gartner to evaluate data science platforms
Gartner vs Forrester Evaluation of Predictive Analytics Platforms
The author discusses Gartner's and Forrester's comparison of predictive analytics platforms along with some other platforms not covered by either company
An Interactive Tool for Evaluating Analytics Platforms
The author illustrates an interactive tool for evaluating predictive analytics platforms
An Interactive Platform Tool for Evaluating Analytics Platforms
Provides a link to an interactive tool for evaluating predictive analytics platforms
Resource provides a link to the Gartner data science platform survey
Module 6: Applications of Machine Learning
Overview of Leading Industry Sectors of Adoption
We discuss which industries are the leading adopters of machine learning technology across the value chain and in what technologies is the money invested
Case Study in Social Networks (Yelp)
The author presents a detailed discussion of how Yelp is using machine learning to get a better understanding of local businesses. We also discuss the corresponding applications Yelp has developed as a result of this new understanding.
Case Study in Entertainment (Netflix)
The author discusses how Netflix is using machine learning to solve the problem of personalized movie recommendations
Case Study in Transportation — Driverless Cars
The author provides a detailed look and several videos of machine learning in action for driverless cars
Applications of Machine Learning in Retail and Sales
A brief presentation how machine learning is being used by Coca Cola and how it can be used to improve sales
Applications of Machine Learning in the Travel Industry
A discussion of how to use machine learning in the travel industry
Applications of Machine Learning in Financial Services
A discussion of how machine learning is being used in financial services and in how it can be used in future applications
Applications of Machine Learning in Healthcare
A discussion of how machine learning is being used in healthcare to supplement the skills of physicians and aid in diagnosis
Future Applications of Machine Learning
A discussion of potential future applications of machine learning and some of their limitations
Module 7: The Technology Stack
The Technology Stack
Introduces the student to the technology that they will need to perform the lab exercises in the remaining courses in the Machine Learning track. The author discusses what is Anaconda, the Spyder IDE, and Jupyter notebooks.
References and Resources
The author provides a quick overview of the top ten Jupyter notebook tutorials for data science and machine learning, and briefly reviews some libraries for machine learning, natural language processing, and plotting and visualization.
Provides instructions for installing the Anaconda distribution
Top 10 Jupyter Notebook Tutorials for Data Science and Machine Learning
Provide links to the top 10 Jupyter notebook tutorials for data science and machine learning
Additional Libraries for Machine Learning, Data Mining and Natural Language Processing, and Plotting and Visualization
Provides links to additional libraries for machine learning, data mining and natural processing, and libraries for plotting and visualization
Sample Python Code for Hierarchical Clustering
The author provides a PDF of sample code for hierarchical clustering to illustrate how to use Jupyter notebooks in conjunction with Python