Experiment Design in Computer Sciences University of Tsukuba
(Course in English) This course is an overview of the basic knowledge needed for performing proper scientific experiments in the field of Computer Sciences. Therefore, this course focuses on three topics: 1- Philosophical discussion about the scientific methods, 2- Conceptual discussion about the design and analysis of Experiments, and 3- Statistical methods for the analysis of experimental data. After finishing this course, the student will be able to consider a research topic in Computer sciences, define what kind of data is necessary for advancing the knowledge in this topic, design an experiment to obtain this data, and evaluating the data in order to draw conclusions from the experment in a rigorous manner. The evaluation method for this course is a short experiment performed in groups, which will be designed, executed, evaluated, presented and reviewed by the students by the end of the course.
The collection and analysis of data through experiments is one of the cornerstones of the scientific method. In this course, we study the general philosophy and methods behind experimentalism: Why do we perform experiments, what is a good/rigorous experiment, how to plan and design a rigorous experiment, and how to perform statistical analysis on experimental data. This course is centered around lectures with plenty of examples and study cases. The students will be invited to apply the techniques studied in this lecture to experiment of their own design.
Knowledge Utilization Skills, Management Skills, Teamwork Skills, International Skills, Research Skills, Ethics
The course will be graded on reports and a Final Exam. The grade proportion is 70% for the reports, and 30% for the exam. The reports require the student to plan, execute, and analyze a simple experiment. This experiment can be about the student's own master research, or about a simple experiment executed during the course. The results of the experiment are to be analyzed and discussed using the techniques learned in this lecture. The report is evaluated on the quality of the experimental design, the correctness of the statistical analysis, and the quality of the discussion of the results. The Final Exam is open book, and covers the entire material of the course.
Introduction - What is science, and what is a scientific experiment? What is rigorous research?Class 2: Statistics review - Point Indicators and Interval Indicators (Random Variables, Means, Variances, Distributions, Confidence Intervals)Class 3: Introduction to Inference Testing -- Hypothesis, Type Errors, Z testing.Class 4: Introduction to Inference Testing -- Comparison Testing, Paired TestingClass 5: Introduction to Inference Testing -- Equality Testing, Non-Parametric TestingClass 6: Lecture Review and Case StudyClass 7: Power Analysis, Sample Size and Sample ChoosingClass 8: Multiple Comparison: Anova and post-hoc testingClass 9: Blocking and Parameter ChoiceClass 10: Lecture Review and Case Study
Online Course Requirement
Sakurai Tetsuya,Aranha, Claus
Lecture in English I expect the students to contribute with questions and discussions during the class. Additionally, I will recommend that the students use data from their own research to prepare the reports. Students who cannot provide their own data will be provided with sample data, however you will learn much more if you can use your own data.
Site for Inquiry
Link to the syllabus provided by the university