Classes meet:

With the explosion of interest in data science, the idea of a "data science pipeline" or

There are multiple courses available to students that focus on data science programing

applications (specifically Python and R) and there are courses focused on analytics which

usually include statistical and machine learning tools. In these courses the data are

the pipeline is ignored; specifically, how should you collect your data?

From a business perspective, knowledge about how to thoughtfully collect data translates

to improved efficiency and reliability. By using a suitable data collection strategy there can

be clear potential time and financial benefits for an organization.

student’s understanding of the complete data science pipeline. At the same time, important

current ideas such as data confidentiality and ethical considerations will be included.

implementation of the different strategies. Given the data science perspective of this

course, this is focused on web enabled approaches. R and/or Python familiarity are

prerequisites for this course, so we will leverage these skills.

At the end of the course, students will have a solid grasp of different data collection

strategies and when and how they can be applied in practice.

requirement. Statistics, through the level of multiple regression is required. This

requirement may be fulfilled with undergraduate courses such as Stat 102, Stat 112, and

MBA courses such as Stat 613/621, or by waiving MBA statistics.

All course materials, including class notes and assignments, will be available on Canvas.

When appropriate, homeworks will be run through Turnitin to determine originality. Late

homeworks are penalized by 25% up to 1 day late and 50% up to two days late.

Homeworks more than two days late will receive a 0.

There will be 5 in-class quizzes. They are closed book. Each quiz has 5 questions (multiple

presented too. We will discuss conditions under which each design might be preferred. The

analysis of a survey that includes survey weights will be discussed.

The principles of experimental design provide the foundations for collecting data in the

most efficient and informative manner. The essential benefits of randomization will be

layout will be described.

across homogeneous subgroups. By blocking, noise can be reduced resulting in more

efficient estimates. Blocking plays the same role as stratification in sampling. The two-way

there are times when they are infeasible. Sometimes because of ethical concerns, and other

times because they are very costly and time consuming. Observational studies lack the

end of the course. Grade queries must be submitted within one week of the homework

solutions being posted.

You can expect that a final score in the 90’s will receive some form of A, 80’s some form of B

percentage of students that will receive a particular grade – everyone could get an A or

everyone could get a C, but that’s unlikely!

There is no formal participation component to the final grade, but questions are strongly

encouraged.