-This is a project that uses the shopping_behavior_updated dataset from kaggle.

-The data set has 18 columns, the major ones including the item purchased, Category, and purchase amount.

Rows/Columns: 3900 rows × 18 columns

Columns:
• **customer_id** (numeric)
• **age** (numeric)
• **gender** (character)
• **item_purchased** (character)
• **category** (character)
• **purchase_amount_usd** (numeric)
• **location** (character)
• **size** (character)
• **color** (character)
• **season** (character)
• **review_rating** (numeric)
• **subscription_status** (character)
• **shipping_type** (character)
• **discount_applied** (character)
• **promo_code_used** (character)
• **previous_purchases** (numeric)
• **payment_method** (character)
• **frequency_of_purchases** (character)

dat <- raw |>
  mutate(
    gender = factor(gender),
    category = factor(category),
    season = factor(season),
    subscription_status = factor(subscription_status, levels = c("No","Yes")),
    discount_applied = factor(discount_applied, levels = c("No","Yes")),
    promo_code_used = factor(promo_code_used, levels = c("No","Yes")),
    shipping_type = factor(shipping_type),
    payment_method = factor(payment_method),
    size = factor(size, levels = c("XS","S","M","L","XL")),
    # Helpful numeric versions
    purchase_amount = `purchase_amount_usd`,
    review_rating = `review_rating`,
    prev_purchases = `previous_purchases`
  ) |>
  filter(!is.na(purchase_amount), purchase_amount > 0)

summary(dat$purchase_amount)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   20.00   39.00   60.00   59.76   81.00  100.00

Why this plot? This bar chart visualizes the distribution of purchases across different product categories. Each bar represents the total number of transactions within a category, allowing for an easy comparison of which types of products are most popular among customers.

Why this plot? This box-and-jitter plot compares purchase amounts across gender groups, giving a clear view of central tendencies, spread, and outliers for each category. The boxplot shows the median (central line) and interquartile range (IQR), representing where the middle 50% of purchase values lie for each gender.The points are densely packed around the median, showing that most customers make mid-range purchases between 50–70 dollars. Both distributions are symmetric and similar in range, meaning gender does not strongly influence purchase amount in this dataset.

Why this plot? The 3D scatter plot effectively captures how age, previous purchase history, and spending amount interact across customers, while also revealing whether gender differences play a role. It helps uncover whether older or more loyal customers tend to spend more, and whether such patterns differ between genders.

Why this plot? The “Seasonal Purchase Share” plot indicates that shopping activity is consistent throughout the year, with only minor variations across seasons.

Commentary:
- Positive coefficient on prev_purchases would suggest more loyal customers tend to spend more per purchase (holding other factors constant).
- Significant effect of discount_appliedYes would imply discounts increase basket size.
- gender coefficients capture average differences vs. the baseline level.
- Check CIs and p-values to discuss significance in your narration.

# Label-dollar formatting helper
fmt_dollar <- function(x) scales::dollar(x, accuracy = 1)

# Convenience plot theme
theme_course <- function() {
  theme_minimal(base_size = 14) +
  theme(panel.grid.minor = element_blank())
}