Academic goals
Analytical techniques of machine learning have proven to be especially useful in predictive areas of different productive sectors, such as tourism, footwear, and industry. How are these methods designed, and how are they instantiated and interpreted?
Learning objectives
This program provides expertise and operational skills on selecting characteristics, predicting numerical variables, regression, predicting categorical variables, classification, patterns.
The following topics are addressed:
1. Association rules models: frequency counting and A priori algorithms.
2. Classification trees with discrete and mixed variables. C4.5 algorithms. Systems of rules and tree parameterization. Confusion matrices and other metrics.
3. Regression trees with discrete and mixed variables. CART algorithms. Systems of equations and trees. Precision metrics.
4. Automatic selection of characteristics: principal component analysis, correlations.
5. Parameterization and outcomes with Weka.
6. Parameterization and outcomes with RStudio.
Facilities and schedule
This program is 40 hours long, 20 of which are given via video conference over the Adobe Connect application. The remaining 20 hours consist of directed activities and materials for students to work on.
Its 20 online classroom hours consist of four 5-hour-long sessions, wherein the faculty present fundamentals about the subject matter.
This program addresses the following topics during the indicated sessions:
1. Association rules models: frequency counting and A priori algorithms. (Sessions 1, 5, 6, and 8)
2. Classification trees with discrete and mixed variables. C4.5 algorithms. Systems of rules and tree parameterization. Confusion matrices and other metrics. (Sessions 1, 2, 5, 6, and 8)
3. Regression trees with discrete and mixed variables. CART algorithms. Systems of equations and trees. Precision metrics. (Sessions 1, 2, 5, 6, and 8)
4. Automatic selection of characteristics: principal component analysis, correlations. (Sessions 1, 2, 5, 6, and 8)
5. Parameterization and outcomes with Weka. (Sessions 3 and 6)
6. Parameterization and outcomes with RStudio. (Sessions 4, 5, 6, 7, and 8)
SESSIONS SCHEDULE
Session 1 (online classroom): April 6, 2020. Introduction to machine learning techniques. A. Rabasa (2.5 hours) and N. Mollá (2.5 hours)
Session 2 (online). Basic problems: readings, tests, and exercises. A. Rabasa (5 h)
Session 3 (online). Introduction to Weka. Video tutorials and exercises. A. Rabasa (5 h)
Session 4 (online classroom): April 21, 2020. Introduction to R (programming language). A. Pérez (5 h)
Session 5 (online) Common R exercise. A. Pérez (2.5 h) and A. Rabasa (2.5 h)
Session 6 (online classroom): April 27, 2020. Review of common R exercise, Weka, and announcement of specific R assignment. A. Pérez-T (2.5 h) and N. Mollá (2.5 h)
Session 7 (online). Specific R exercise, doubts, and follow-up. A. Pérez-T (2.5 h) and N. Mollá (2.5 h)
Session 8 (online classroom): May 5, 2020. Presentations and discussion about specific R exercises. A. Pérez-T (1.5 h), N. Mollá (1.5 h), A. Pérez (1 h) and A. Rabasa (1 h)