05_workbook.Rmd

---
title: "Data Viz and Regression in R"
author: "Aaron R. Williams"
output:
  html_document:
    code_folding: show
    toc: TRUE
    toc_float: TRUE
editor_options:
  chunk_output_type: console
---

```{r}
library(ggplot2)
library(tidyverse)
library(broom)
library(modelr)


theme_set(theme_minimal())

```

## Exercise 1

```{r}
# sample 300 observations and set ordinal factors to nominal
set.seed(20200622)

diamonds <- diamonds %>%
  slice_sample(n = 300) %>%
  mutate(across(where(is.factor), .fns = as.character))

# estimate a multiple linear regression model
diamonds_model1 <- lm(formula = price ~ carat + cut, data = diamonds)

## YOUR WORK GOES HERE
# estimate a multiple linear regression model

class(diamonds_model1)

augment(diamonds_model1) %>%
  ggplot(aes(x = .fitted, y = .resid)) +
  geom_point(alpha = 0.2) +
  geom_smooth()


 ```

## Exercise 2

```{r}
# fit and tidy model 1
diamonds_model1 <- lm(formula = price ~ carat + cut, data = diamonds)

diamonds_model1_coefs <- tidy(
  diamonds_model1, 
  conf.int = TRUE,
  conf.level = 0.95
) 

# fit and tidy model 2

diamonds_model2 <- lm(formula = price ~ carat + cut + x, data = diamonds)

diamonds_model2_coefs <- tidy(
  diamonds_model2, 
  conf.int = TRUE,
  conf.level = 0.95
)

## YOUR WORK GOES HERE

# combine the model results
models_coefs <- bind_rows(
  `model1` = diamonds_model1_coefs,
  `model2` = diamonds_model2_coefs,
  .id = "model"
)

# visualize

## YOUR WORK GOES HERE

models_coefs %>%
  ggplot(aes(x = estimate,
             y = term,
             xmin = conf.high,
             xmax = conf.high,
             color = model)) +
  geom_vline(xintercept = 0) +
  geom_pointrange(position = position_dodge(width = 0.5)) +
  scale_x_continuous(limits = c(-15000, 15000),
                     labels = scales::dollar)

```

## Exercise 3

```{r}
# estimate a linear model for each country
many_models <- diamonds %>%
  group_by() %>% ## YOUR WORK GOES HERE
  nest(data = c()) %>% ## YOUR WORK GOES HERE
  mutate(
    model = map(
      .x = data, 
      .f = ~glance(lm(formula = , data = .)) ## YOUR WORK GOES HERE
    )
  ) %>%
  ungroup()
    
# extract r^2 from each model

## YOUR WORK GOES HERE

# plot

## YOUR WORK GOES HERE

```

## Exercise 4

```{r}
# create a binary outcome variable
cars <- cars %>%
  mutate(crash = as.numeric(dist > 25))

# fit a linear probability model
cars_lm <- ## YOUR WORK GOES HERE

# fit a logistic regression model
cars_glm <- ## YOUR WORK GOES HERE

# add conditional probabilities for both models
models <- data_grid(cars, speed = seq_range(speed, 1000)) %>%
  add_predictions(cars_lm, var = "lm") %>%
  add_predictions(cars_glm, type = "response", var = "glm")

# visualize

## YOUR WORK GOES HERE

```