Homework 5

Generative AI for Data Analysis

Author

Byeong-Hak Choe

Published

December 2, 2025

Modified

December 2, 2025


Question 1. Choice of Generative AI Tool

What generative AI tool have you used for this homework assignment?


Question 2. Python Visualization with the seaborn Library

seaborn
  • seaborn is a Python data visualization library that provides a high-level, elegant interface for creating informative and attractive graphics.
    • You can think of it as the Python counterpart to R’s ggplot2: it emphasizes clear defaults, aesthetic color palettes, and concise syntax for complex visualizations.
    • Because of its intuitive design and visually appealing output, I recommend using seaborn as the default choice for visualization in Python.
  • Provide your conversation with generative AI to do the following tasks:
    • Translate the following R ggplot code into Python seaborn code to generate a scatter plot showing the relationship between β€œsales” and β€œprice” using the CSV file, http://bcdanl.github.io/data/dominick_oj_na.csv.
    • Make a step-by-step comparison between the Python code and the R code to understand how each part corresponds to the other.
library(tidyverse)

oj <- read_csv("http://bcdanl.github.io/data/dominick_oj_na.csv")

ggplot(data = oj, 
       mapping = aes(x = price, y = sales,
                     color = brand)) +
  geom_point(alpha = .3) + 
  geom_smooth(method = "lm") +
  labs(title = "Scatter Plot of Sales vs. Price",
       x = "Price",
       y = "Sales")


Question 2. Python Data Analysis with the pandas Library

pandas
  • pandas is Python’s primary library for data manipulation and analysis, offering powerful tools for working with tabular data.
    • You can think of it as the Python counterpart to R’s dplyr (tidyverse): it provides clear, expressive functions for filtering, transforming, summarizing, and reshaping data.
  • Because of its intuitive syntax and flexible data structures β€” DataFrame (similar to a data.frame in R) and Series (similar to a vector in R) β€” pandas is the default foundation for most data analysis workflows in Python.

  • Provide your conversation with a generative AI tool to complete the following tasks:

    • Translate the R dplyr code below into equivalent Python code using pandas to perform the same data-manipulation steps on the CSV file
      http://bcdanl.github.io/data/dominick_oj_na.csv.

    • Make a step-by-step comparison between the Python code and the R code, explaining how each part of the dplyr pipeline corresponds to the equivalent operation in pandas.

library(tidyverse)

oj <- read_csv("http://bcdanl.github.io/data/dominick_oj_na.csv")

oj |>
  select(brand, price, sales) |>
  filter(price > 1.5) |> 
  mutate(log_sales = log(sales)) |>
  arrange(desc(price))


Question 4. Debugging R Code with Generative AI

  • Provide your conversation with generative AI to debug the following code:
    • Explain to the generative AI the error message you received when running the code below:
oj |> 
  counting(brand, ad_status)


Question 5. Data Transformation with Generative AI Assistance

  • Provide your conversation with generative AI for adding a new variable, revenue, to the oj data frame.
    • The revenue variable should be computed as the product of sales and price to provide information about the weekly revenue for each orange juice brand.
library(tidyverse)
oj <- read_csv("http://bcdanl.github.io/data/dominick_oj_na.csv")


Question 6. Understanding R ggplot and dplyr Code with Generative AI

  • Provide your conversation with a generative AI tool to complete the following task:
    • In the code below, add a comment (#) with a brief explanation for each line of the R code below to show your understanding of what the code is doing.
library(tidyverse)
library(hrbrthemes)
library(ggthemes)
library(grid)
library(scales)

events_raw <- read_csv("https://bcdanl.github.io/data/time-series-US-cost-1980-2024.csv",
                       comment = "#") |> 
  select(-matches("upper|lower", ignore.case = TRUE))

events_counts <- events_raw |>
  select(Year, matches("Count$"))  

events_long <- events_counts |>
  pivot_longer(
    cols      = -Year,
    names_to  = "hazard",
    values_to = "count"
  ) |>
  mutate(
    hazard = str_remove(hazard, " Count$"),
    hazard = factor(
      hazard,
      levels = c(
        "Drought",
        "Tropical Cyclone",
        "Flooding",
        "Freeze",
        "Severe Storm",
        "Winter Storm",
        "Wildfire"
      )
    )
  ) |>
  filter(!is.na(hazard))

cost_df <- events_raw |>
  select(Year, matches("All Disasters.*Cost", ignore.case = TRUE)) |>
  rename(cost_billion = 2) 

plot_df <- events_long |>
  left_join(cost_df, by = "Year")


max_count <- max(plot_df$count, na.rm = TRUE)
max_cost  <- max(plot_df$cost_billion, na.rm = TRUE)
cost_scale <- max_count / max_cost



hazard_cols <- c(
  "Drought"          = "#E69F00",  # orange
  "Tropical Cyclone" = "#007F7F",  # green
  "Flooding"         = "#56B4E9",  # light blue
  "Freeze"           = "#0072B2",  # dark blue
  "Severe Storm"     = "#4F6D8A",  # purple
  "Winter Storm"     = "#999999",  # gray
  "Wildfire"         = "#D55E00"   # red
)

ggplot(plot_df, aes(x = Year)) +
  geom_area(
    aes(y = count, fill = hazard),
    position = "stack",
    color = NA
  ) +
  geom_smooth(
    aes(y = cost_billion * cost_scale),
    linewidth = 3,
    color = "#2c2e2f",
    se = FALSE
  ) +
  geom_label(
    data = tibble(
      Year = 2004,
      y    = max_count * 0.95,
      lab  = "Yearly Trend of\n Billion-Dollar\nDisaster Cost"
    ),
    aes(x = Year, y = y, label = lab),
    inherit.aes = FALSE,
    hjust = 0.5,
    vjust = 0.5,
    size  = 4,
    fontface = "bold",
    label.size = .5,      
    label.r = unit(0.15, "lines"),
    label.padding = unit(0.5, "lines")
  ) +
  annotate(
    "segment",
    x = 2004,
    xend = 2004, 
    y = 15,
    yend = 3,
    color = "black",
    linewidth = 0.5,
    arrow = arrow(length = unit(0.25, "cm"), type = "closed")
  ) +
  scale_fill_manual(
    values = hazard_cols,
    name   = ""
  ) +
  scale_y_continuous(
    name = "Number of Events",
    breaks = seq(0,30,10),
    limits = c(0,30),
    sec.axis = sec_axis(
      ~ . / cost_scale, 
      name = "Cost in billions",
      labels = label_dollar(prefix = "$")
    )
  ) +
  scale_x_continuous(
    breaks = seq(1980,2025,5)
  ) +
  labs(
    x = NULL,
    fill = "",
    title = "U.S. Billion-Dollar Disasters: Events and Costs 1980-2024 (CPI-Adjusted)",
    caption = "Source: National Oceanic and Atmospheric Administration (2024)"
  ) +
  theme_ipsum() +
  theme(
    plot.title = element_text(hjust = 0.5,
                              size = rel(1.75)),
    panel.grid.minor = element_blank(),
    axis.title.y = element_text(size = rel(1.5),
                                margin = margin(0,12,0,0)),
    axis.title.y.right = element_text(size = rel(1.12),
                                      margin = margin(0,0,0,12)),
    legend.position = "top",
    legend.box = "horizontal",
    legend.direction = "horizontal"
  ) +
  guides(
    fill = guide_legend(
      title.position = "top",
      label.position = "bottom",
      keywidth = unit(4.75, "lines"),
      nrow = 1
    )
  )

References


Back to top