Week_1_Notes.Rmd

---
title: "Week 1 Notes"
output:
  html_document:
    df_print: paged
---

## Notes for Intoductory Class - Week 1

#### Personal background on statistics and this book
  - little background in statistics
  - frustrated with PCMs and the lack of flexibility
  - Didn't realize all the options out there!

#### Multilevel modeling
  - Allows us to create generative models of our data, or the joint probabilities of our variables
  - There are natural structures in our data that can't be ignored, like taking the mean of a species

#### Bayesian Approach
  - Offers the most logical and practical way to model our data
  - Conducive to regularization, missing data, etc,
  - Probabilities offer an intuitive way to explain and understand our systems
  - There is a principled workflow
  - Many tools are tighly integrated with other data science tools for easy visualization, data wrangling, etc.
  
#### This Seminar Course
  - In addition to the text, using the 2022 lectures and the rethinking in brms book
  - Newest version of the course focus heavily on scientific models, causal inference, and DAGs. New to me!
  - Explain expectations -> Mark?
  - Principles Bayesian Workflow: Scientific Model -> Statistical Model -> Prior Predictive Simulation -> Model Fitting -> Predictive Accuracy/Model Comparison -> Posterior Predictive Simulation -> Data Visualization and Reporting
  
#### Tools
  - Rethinking package, and brms package. If you are not wedded to base R, I strongly suggest learning the tidyverse!!!!!
  - RStudio, RMarkdown, knitting
  - Scripting, Software engineering by amateurs. Important to document work, and make it reproducible! And Keep track of it! RMarkdown makes this really easy. 

```{r}
pacman::p_load(tidyverse, brms)
```

```{r}
m1 <- brm(mpg ~ hp, data = mtcars, refresh = 0)

loo(m1)

plot(conditional_effects(m1), points = T)

pp_check(m1, ndraws = 200)
```