Discussion 4

Dataset

Advertising Dataset (Kaggle): https://www.kaggle.com/datasets/ashydv/advertising-dataset

Each observation is a market, with the amount spent on TV, radio, and newspaper advertising ($ thousands) and product sales (thousands of units).

ads <- read.csv("https://raw.githubusercontent.com/selva86/datasets/master/Advertising.csv")
ads <- ads[, -1]

Sample Size

set.seed(123)
sample_ads <- ads[sample(nrow(ads), nrow(ads) / 2), ]

nrow(ads)

## [1] 200

nrow(sample_ads)

## [1] 100

The original dataset has 200 observations. I used a random sample of 100 observations (half) for my analysis.

EDA 1: Summary Statistics and Histogram of Sales

summary(sample_ads$sales)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    5.30   10.25   12.65   13.90   17.32   26.20

sd(sample_ads$sales)

## [1] 4.861288

hist(sample_ads$sales,
     breaks = seq(0, 30, 2.5),
     col = "steelblue",
     main = "Distribution of Sales",
     xlab = "Sales (thousands of units)")

Sales range from 5.3 to 26.2, with a mean of 13.90, a median of 12.65, and a standard deviation of 4.86. The distribution is slightly right-skewed, with most markets selling between 10 and 15 thousand units.

EDA 2: Correlation and Scatterplot

round(cor(sample_ads), 3)

##               TV radio newspaper sales
## TV         1.000 0.000    -0.078 0.785
## radio      0.000 1.000     0.346 0.524
## newspaper -0.078 0.346     1.000 0.068
## sales      0.785 0.524     0.068 1.000

plot(sample_ads$TV, sample_ads$sales,
     pch = 19, col = "darkorange",
     main = "TV Ad Budget vs. Sales",
     xlab = "TV ad budget ($ thousands)",
     ylab = "Sales (thousands of units)")
abline(lm(sales ~ TV, data = sample_ads), col = "red", lwd = 2)

Sales have a strong positive correlation with TV spending (r = 0.785), a moderate correlation with radio (r = 0.524), and almost no correlation with newspaper (r = 0.068).

What I Learned