MAA 540 Machine Intelligence (Spring 2022)
Dale Lehman
Dale.Lehman@loras.edu
AI
(Artificial Intelligence) and Machine Learning, often used
interchangeably, are the latest buzzword to have usurped Big Data.
The promises, and dangers, are large indeed. Algorithms
have outperformed doctors, teachers, Jeopardy contestants, chess
masters, and many other trained professionals. They may soon take
over driving and battlefields. They have proven to be more
efficient and more accurate than humans, and often less biased as well
(though their biases exist and may be harder to identify). The
variety of algorithms that can be used to base decisions on data grows
daily (mostly some variety of decision trees, neural networks,
convolutionalnetworks, boosted trees, random forests, and regression
models).
As these modeling techniques grow more powerful, a new class of
software tools has developed - referred to as Auto ML (automated
machine learning). For better or (and) worse, analysis
increasingly relies on computer processing of data, and the role of the
data scientist/analyst relies on learning to effectively work with
computers and humans each doing what they are best at. This
course will explore these tools, their
requirements, capabilities, and how they can be deployed in
organizational settings. The cousrse will utilize JMP, Alteryx, KNIME,
and
DataRobot.
Grading: There will be homework
assignments and a group project. Homework will comprise 2/3 of
the course grade and the group project the other 1/3. Homework
can be resubmitted (all or partially) for up to half of the orginal
grade, after it was due and has been gone over in class. The
grading rubric is:
A: demonstrate a command of the tools
B: control of the tools, but too many omissions or errors to call it command.
C: effort is clear, but the tools are controlling you rather than vice versa.
D: effort is questionable.
F: lack of effort has been demonstrated.
Textbook: textbook materials on eLearn.
Brief Outline
Classification Modeling: data preperation, techniques, validation, model evaluation
Logistic
Regression, Decision Trees, Boosting, Random Forests, Neural Networks,
Naive Bayes, K Nearest Neighbors, Support Vector Machines
Accessing and Using Data
Querying, joining, weighting, and documenting work flow
Group Project (TBD)