Skip to content

Latest commit

 

History

History
155 lines (103 loc) · 5.28 KB

ideas.md

File metadata and controls

155 lines (103 loc) · 5.28 KB

2021 autumn

  • Ask students to solve problems and then ask for answers during the meetings.
  • Collect questions about topics somehow.
  • At the end of final meeting ask for feedback about the course.
  • At the end of each meeting first one exercise is demonstrated and then students individually or in groups solve on other data sets.
  • Find out and write down classrooms!
  • Update slides to notes
  • Explain how to use Jamovi for research report
  • Add to notes a list of things that need to be interpreted for each method.
  • OLS and multiple OLS are overlapping method in terms of research report. Only score one of these.

2021 spring

  • Use Jamovi, for R +20%
  • datasets
    • Different data for same method to illustrate different results.
  • show regression elements on plots
  • come up with the types of variables that the data needs to have.
  • Use Jamovi textbook + additional source for GLM
  • Do an initial test to understand present skills and expectations.
    • Use the results later to demonstrate some methods
  • Show common transformations, e.g. normalizing, scaling.
  • Course is structured by methods: new method in each meeting.
    • Material for each method involves:
      • theory
      • implementation in software
      • interpretation of results
      • considerations
  • Do a lot of different exesrcises with different data during meetings.
  • Talk about why even do research and what can statistics show, some philosophy on this first.

Cluster analysis: show matrix step by step for 2 variables. Introduce machine learning.

Theory vs empirics

First time: install and try out RStudio and Jamovi, read some introductory materials.

15 minutes of explaining at a time.

Classification and regression trees.

Instructions on data set. If can't do an analysis then explain why.

On Wednesday 21-05-05

Before

  • Find data for examples on categorical.
  • Add comics to slides.

Start with descriptive statistics in Jamovi and then R.

Teach R as well. Explain how last time wasn't successful. We will not be covering the intricacies if R. But learn yourself at http://adv-r.had.co.nz/ (Data structures, Subsetting, Vocabulary).

Ask if have found data. If not, then give it.

Discuss the book by Navarro and ... (2018). Students should read it. Notes are only condensed version of it to get some quick information.

Flipped classroom. Theory learning at home, meetings for: answering questions, summarizing and explaining, implementation in practice.

As little monologue as possible.

Focus on methods and application and not theory.

Idea: teach methods, not theory. Bloom's Taxonomy. Traditional methods: why.

Research project. Again!?

Cameras and microphones on. We also practice soft skills: English, communication and presentation

Not going to introduce myself. Search from Google if interested.

Read materials to learn how to learn.

Data. Can have different data for each meeting. As long as it's different from other students. Not limited to agriculture and food production.

The book helps to organise (only to some extent summarise) the necessary knowledge.

Topics

Focus on neccessary tools.

  1. Descriptive
    1. Scales of measurement (LSJ 16)
    2. Measures of central tendency (LSJ 4.1)
    3. Measures of variability (LSJ 4.2, , 4.3)
    4. Plots (scatterplot, histogram, boxplot) (LSJ 5)
    5. Suggestions for plotting
  2. Samples, populations (LSJ 8). Hypothesis testing (LSJ 9)
    1. Categorical data analysis (LSJ ch 10)
    2. Odds ratios.
  3. Parametric tests, non-parametric tests (LSJ 11)
    1. Multiple testing problem
  4. Analysis of variance (LSJ 13, 14)
    1. Tukey's HSD
  5. Correlation (LSJ 12.1)
    1. Heatmaps
    2. Causality
  6. OLS regression (RBook 10.1, 10.13)
    1. Model checking (RBook 9.13), BLUE
    2. Model criticisism (RBook 9.12)
  7. OLS transformations, interactions, factor variables
    1. Model formulae (RBook 9.10)
    2. Box-Cox
    3. Overfitting
  8. GLM regression (logit/probit) (RBook 13)
    1. Diagnostic tests
    2. Odds ratios
  9. PCA (LSJ 15.2, RBook 25.1)
  10. Factor analysis (LSJ 15.1, RBook 25.2). Variamax, oblimin
  11. Hierarchical clustering (RBook 25.4), K-means clustering (RBook 25.3)
  12. Presentations

2019 autumn

  • Overview of simple things, i.e. elements, vectors, data frame. Creating and subsetting them but nothing more. Also logical operators, getting an overview.

  • Package management.

  • Some tips on using computer and typing.

  • Show how to use some example data, e.g. from agridat().

  • Explain data import from .xlsx.

  • Don't go beyond simple things, stick with what is provided by dplyr. Perhaps coupled with base R equivalents. Teach View().

  • Provide visualizations of everything possible.

  • Give home reading each time and test it each time.

  • Teach just regression, only brief overview of hypothesis testing. Simple LS, multivariable LS, logit/probit, ordered, multinomial, count. Also, static panel data.

  • Stochastic frontier analysis?

  • Explain BLUE, incl. why shouldn't some models panel data.

  • Transformed data, e.g. log-linear.

  • Multiple testing problem.

  • Explain well how one endogenous variable may affect another. Also,

    interactions.

  • Final task with some personally chosen data, perhaps homework.

  • Tests on each meeting, home assignment on use of skills instead of test.

  • When using a command, also make sure it was effective afterwards.