One of the principal responsibilities of a data scientist is to make reliable predictions based on data. When the amount of data available is enormous, it helps if some of the analysis can be automated. Machine learning is a way of identifying patterns in data and using them to automatically make predictions or decisions. In this data science course, you will learn basic concepts and elements of machine learning.
Individual course: $199
Course duration: 6 weeks
Time commitment: 4-6 hours each week
Experience level: Introductory
Learning partner: University of California, Berkeley
Course type: Self-paced on your time
Subjects: Computer science, data analysis, statistics, and IT
The two main methods of machine learning you will focus on are regression and classification. Regression is used when you seek to predict a numerical quantity. Classification is used when you try to predict a category (e.g., given information about a financial transaction, predict whether it is fraudulent or legitimate).
For regression, you will learn how to measure the correlation between two variables and compute a best-fit line for making predictions when the underlying relationship is linear. The course will also teach you how to quantify the uncertainty in your prediction using the bootstrap method. These techniques will be motivated by a wide range of examples.
For classification, you will learn the k-nearest neighbor classification algorithm, learn how to measure the effectiveness of your classifier, and apply it to real-world tasks including medical diagnoses and predicting genres of movies.
The course will highlight the assumptions underlying the techniques and will provide ways to assess whether those assumptions are good. It will also point out pitfalls that lead to overly optimistic or inaccurate predictions.
Topics of study
Fundamental concepts of machine learning
Linear regression, correlation, and the phenomenon of regression to the mean
Classification using the k-nearest neighbors algorithm
How to compare and evaluate the accuracy of machine learning models
Basic probability and Bayes’ theorem
About the University of California, Berkeley
The University of California, Berkeley was chartered in 1868, and its flagship campus — envisioned as a "City of Learning" — was established at Berkeley, on San Francisco Bay. Berkeley faculty consists of 1,582 full-time and 500 part-time faculty members dispersed among more than 130 academic departments and more than 80 interdisciplinary research units. Berkeley alumni have received 28 Nobel prizes, and there are eight Nobel laureates, 32 MacArthur Fellows, and four Pulitzer Prize winners among the current faculty.
In September 2012, to mark Berkeley's commitment to innovation in teaching and learning, The Berkeley Resource Center for Online Education (BRCOE) was formed. The Center is a resource hub and an operational catalyst for all internal campus-wide and external resources to advise, coordinate, and facilitate the university’s online education initiatives, ranging from credit and non-credit courses, to online degree programs and MOOC projects, including the MOOCLab initiative.
BRCOE's new MOOCLab is a three-year research initiative to fund and develop Massive Open Online Courses (MOOCs) as vehicles for pedagogical research in online education.
Berkeley is also working with edX to develop and foster adoption of Small Private Online Courses (SPOCs) on campuses around the world. SPOCs are designed to supplement and enhance the learning experience of on-campus students, while providing local faculty an opportunity for more interactive activities and more time for “high-touch” pedagogy.