Overview

For this dataset, I’ll be using the first one posted by Waheeb and it represents sales data for different product lines based on a specific date. This data has one row per date and includes the following columns:

The last three columns are dollar values.

product_data <- read.csv("https://raw.githubusercontent.com/addsding/data607/main/project2/product-lines.csv")
head(product_data)

Our goal is to flatten this table to be one row per date and product line number combination before beginning analysis.

Tidying the Data

To clean this data frame, we’ll be pivoting it.

product_data_pivot <- pivot_longer(
  product_data, cols=2:4, names_to="product_line", values_to="sales"
  )
head(product_data_pivot)

The pivot has worked! There are however periods instead of spaces in the product_line column so we’ll replace those. Also, the Date column also should be changed into an actual date.

product_data_pivot$product_line <- gsub("\\.", " ", product_data_pivot$product_line)

product_data_pivot$Date <- as.Date(product_data_pivot$Date,
  format = "%m/%d/%y")

head(product_data_pivot)

Looks good, time to analyze!

Analysis

To begin, we can find the stats for sales for each product line.

sales <- product_data_pivot |> 
  group_by(product_line) |>
  summarise(mean_sales = mean(sales),
            median_sales = median(sales),
            min_sales = min(sales),
            max_sales = max(sales),
            .groups = 'drop')

sales

On average, product line 3 seems to be doing the best as on average, it has sales of $2,500+. Product line 2 isn’t that far behind at $2,450, while product line 1 seems to trail behind at only $1,700. We can see though that product line 3 has a pretty large range of sales from $850 to $5,000 – could this be due to seasonality?

product_time_series <- ggplot(product_data_pivot, aes(x=Date, y=sales, color=product_line)) +
  geom_line() + 
  xlab("")
product_time_series

When looking at this data over time, it tells a very different story. Product line 2 seems to be growing a lot while product line 3 has not been doing so well, really tanking in sales for the first quarter of the year. Product line 1 was relatively consistent after a pretty huge drop from January to February, however seems to be bouncing back as of June.

Conclusion

Overall, this data was relatively simple to clean and the findings were pretty straight-forward. Two piece of informations that I think are super important here though is supply as well as overall price of each product line – we can’t really compare the performance of each product line without knowing how much each unit costs as well as how much was actually produced. If for example, product line 1 is priced at a lower tier and had less units made, its performance would actually be more impressive if product line 3 was over-produced and was at a much higher price point. To continue with this analysis, we’d definitely need these data points at the very least to fully gage how well each product line is performing.