Loading R packages for Homework Assignment 3

library(tidyverse)
library(lubridate)
library(ggthemes)



Question 1

Read the data file, NY_school_enrollment_socioecon.csv, as the data.frame object with the name, NY_pincp, using (1) the read_csv() function and (2) its URL, https://bcdanl.github.io/data/NY_school_enrollment_socioecon.csv.

url <- 'https://bcdanl.github.io/data/NY_school_enrollment_socioecon.csv'
NY_school_enrollment_socioecon <- read_csv(url)



For description of variables in NY_school_enrollment_socioecon, refer to the file, ny_school_enrollment_socioecon_description.zip, which is in the Files section in our Canvas web-page. (I recommend you to extract the zip file, and then read the file, ny_school_enrollment_socioecon_description.csv, using Excel or Numbers.)



Q1a

Look up the meaning of variable pincp from the data.frame, NY_pincp. Create the data frame, NY_pincp, which has only the following four variables from NY_school_enrollment_socioecon:

  • fips
  • year
  • county_name
  • pincp
NY_pincp <- NY_school_enrollment_socioecon %>% 
  select(fips:pincp)

NY_pincp



Q1b

Create the data frame, NY_pincp_wide, which has one observation for each county (62 observations in total) and the following eight variables:

  • fips: ID number for county
  • county_name: County Name
  • pincp2015: Annual personal income in year 2015
  • pincp2016: Annual personal income in year 2016
  • pincp2017: Annual personal income in year 2017
  • pincp2018: Annual personal income in year 2018
  • pincp2019: Annual personal income in year 2019
  • pincp2020: Annual personal income in year 2020
NY_pincp_wide <- NY_pincp %>% 
  pivot_wider(names_from = year,
              values_from = pincp,
              names_prefix = 'pincp')



Q1c

Use the data.frame NY_pincp_wide from Q1b to create the data frame, NY_pincp_long, which has six observations for each county (372 observations in total) and the following four variables:

  • fips: ID number for county
  • year: integer variable of year (e.g., The possible values of year are 2015, 2016, 2017, 2018, 2019, and 2020.)
  • county_name: County Name
  • pincp: Annual personal income
# (1) using separate()
NY_pincp_long <- NY_pincp_wide %>% 
  pivot_longer( cols = pincp2015:pincp2020,
                names_to = "year",
                values_to = "pincp") %>% 
  separate(year, into = c("dum", "year"), 
           sep = "pincp", convert = T) %>% 
  select(-dum)

# () using str_replace()
NY_pincp_long <- NY_pincp_wide %>% 
  pivot_longer( cols = pincp2015:pincp2020,
                names_to = "year",
                values_to = "pincp") %>% 
  mutate( year = str_replace(year, "pincp", ""),
          year = as.integer(year) )



Q1d

Create the data.frame, NY_pincp_geo, by join the two data.frames, NY_county_geo and NY_pincp.

NY_county_geo <- read_csv('https://bcdanl.github.io/data/NY_county_geo.csv')

The data.frame, NY_pincp_geo, must include all the observations and variables in the data.frame, NY_county_geo.

NY_pincp_geo <- NY_county_geo %>% 
  left_join(NY_pincp, by = c("FIPS" = "fips"))



Q1e

In the following ggplot code with geom_polygon(), replace the blanks ([?]) with the appropriate object to draw a yearly county map of pincp.

# ggplot(data = [?]) +
#   geom_polygon(mapping = aes(x = [?], y = [?], group = group, 
#                              fill = [?] ) ,
#                color = "grey", size = 0.1) +
#   facet_wrap( [?] ) +
#   labs( title = "Income in NY State" ) +
#   coord_map("bonne", parameters = 41.6) +  # for better aspect ratio
#   scale_fill_gradient2( #  mapping pincp rate to color level
#   low = 'red', 
#   high = 'blue', 
#   na.value = "grey50",
#   midpoint = quantile(
#     NY_school_enrollment_socioecon$pincp, 50)) +
#   theme_map() +    # a better theme for map drawing
#   theme(legend.position='top') 
ggplot(data = NY_pincp_geo) +
  geom_polygon(mapping = aes(x = long, y = lat, group = group, 
                             fill = pincp ) ,
               color = "grey", size = 0.1) +
  facet_wrap( ~ year ) +
  labs( title = "Average Income in NY Counties, 2015-2020",
        caption = "Source: Breau of Labor Statistics") +
  coord_map("bonne", parameters = 41.6) +  # for better aspect ratio
  scale_fill_gradient2( #  mapping pincp values to color levels
    low = 'red', 
    high = 'blue', 
    na.value = "grey50",
    midpoint = quantile(
      NY_school_enrollment_socioecon$pincp, .50),
    name = "Average Income") +
  theme_map() +    # a better theme for map drawing
  theme(legend.position= "bottom",
        plot.title = element_text(hjust = .5, size = 16,
                                  color = "purple"),
        strip.background = element_rect(color = 'white', 
                                        fill = 'white'))




Question 2

Read the data file, tech_stock.csv, as the data.frame object with the name, tech_stock, using (1) the read_csv() function and (2) its URL, https://bcdanl.github.io/data/tech_stock.

tech_stock <- read_csv("https://bcdanl.github.io/data/tech_stock.csv")
tech_stock



Q2a

Describe the daily trend of Close for each company since 2010 in one ggplot.

ggplot(filter(tech_stock, Date > ymd("2010-01-01")),
       aes(x = Date, y= Close,
           color = company)) +
  geom_line() +
  hrbrthemes::theme_ipsum()

ggplot(filter(tech_stock, Date > ymd("2010-01-01")),
       aes(x = Date, y= Close,
           color = fct_reorder2(company, Date, Close))) +
  geom_line() + 
  hrbrthemes::theme_ipsum()



Q2b

  • Describe the relationship between log(Close) and log(Volume) for the companies in tech_stock.
p0 <- ggplot(tech_stock , 
             aes(x = log(Close), y = log(Volume)) )
p0 + geom_point(alpha = .05,
                shape = 1,
                aes(color = company)) + geom_smooth(method = lm) +
  guides(colour = guide_legend(override.aes = list(alpha=1)))



Q2c

Does the relationship between log(Close) and log(Volume) vary by companies and by the time periods?

# all years
p0 <- ggplot(tech_stock , 
             aes(x = log(Close), y = log(Volume),
                 color = company) )
p0 + geom_point(alpha = .05,
                shape = 1) + geom_smooth(method = lm) +
  facet_grid(.~company) +
  theme_minimal()+
  labs(title = "All years") +
  theme_bw() +
  guides(color = "none")

# since 2017
tech_stock2 <- tech_stock %>% 
  filter(Date >= ymd("2017-01-01")) %>% 
  mutate(year = year(Date))

p0 <- ggplot(tech_stock2 , 
             aes(x = log(Close), y = log(Volume),
                 color = company) )
p0 + geom_point(alpha = .075,
                shape = 1) + 
  geom_smooth(method = lm, se = F, size = .75) + 
  facet_grid(year~company, scales = "free_x") +
  labs(title = "Since 2017") +
  theme_bw() +
  guides(color = "none")