Week 7: Apply it to your data 6

Introduction

Questions

Variation

Visualizing distributions

ggplot(data = data) +
  geom_bar(mapping = aes(x = year))

data %>%
    count(year)

## # A tibble: 21 × 2
##     year     n
##    <dbl> <int>
##  1  1930    18
##  2  1934    17
##  3  1938    18
##  4  1950    22
##  5  1954    26
##  6  1958    35
##  7  1962    32
##  8  1966    32
##  9  1970    32
## 10  1974    38
## # ℹ 11 more rows

ggplot(data = data) +
  geom_histogram(mapping = aes(x = home_score))

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

ggplot(data = data, mapping = aes(x = year, colour = outcome)) + 
  geom_freqpoly()

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Typical values

data %>%
    
    filter(year > 2000) %>%
    
    ggplot(aes(x = year)) +
    geom_histogram(binwidth = 1)

Unusual values

data %>%
    ggplot(aes(home_score)) +
    geom_histogram()

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Missing Values

data %>%
    mutate(home_score = ifelse(home_score > 3, NA, home_score)) %>%
    
    ggplot(aes(x = home_team, y = home_score)) + 
    theme(axis.text.x = element_text(angle = 90, hjust = 1)) +
    geom_point() 

## Warning: Removed 86 rows containing missing values (`geom_point()`).

Covariation

A categorical and continuous variable

data %>%
    ggplot(aes(x = outcome, y = away_score)) +
    geom_boxplot()

Two categorical variables

data %>%
    count(outcome, home_team) %>%
    ggplot(aes(x = outcome, y = home_team, fill = n)) + 
    geom_tile()

Two continous variables

data %>%
    ggplot(aes(x = away_score, y = home_score)) +
    geom_hex()

Patterns and models

library(modelr)

mod <- lm(log(away_score + 1) ~ log(home_score + 1), data = data)

data4 <- data %>%
    add_residuals(mod) %>%
    mutate(resid = exp(resid))

data4 %>%
    ggplot(aes(x = home_score, resid)) +
    geom_point()