Exploratory Data Analysis

Section 9.4 Exploratory Data Analysis

Before we predict if a person defaults or not, it is good to find out more about the default variable, answering some questions like:

🔗

How many people have defaulted?
🔗

🔗
What percentage of people have defaulted?
🔗

🔗
What are the odds that someone will default?
🔗

🔗
What are the preliminary relationships between default and the other variables?
🔗

🔗

table(cc_data$default)

No  Yes 
9677  323

# Proportion of people who default
p <- mean(cc_data$default == "Yes")
p*100

[1] 3.33

odds <- p / (1 - p)

# Below provides us with the default rate
cat("Default rate:", round(p*100,2), "%\n")

Default rate: 3.33 %

# Below provides us with the odds of someone defaulting
cat("Odds of defaulting:", round(odds,4), "\n")

Odds of defaulting: 0.0344

# Below provides us with the odds ratio
cat("Equivalent odds ratio (1 in every", round(1/odds,0), ")\n")

Equivalent odds ratio (1 in every 29 )

# Cross-tab between student status and default
cc_data |> count(default, student) |> arrange(desc(n))

  default student    n
1      No      No 6850
2      No     Yes 2817
3     Yes      No  206
4     Yes     Yes  127

# Quick descriptive stats
library(mosaic)

favstats(income ~ default, data = cc_data)

  default       min       Q1   median       Q3      max     mean       sd    n
1      No  771.9677 21405.06 34589.49 43823.76 73554.23 33566.17 13318.25 9667
2     Yes 9663.7882 19027.51 31515.34 43067.33 66466.46 32089.15 13804.22  333
  missing
1       0
2       0

favstats(balance ~ default, data = cc_data)

  default      min        Q1    median       Q3      max      mean       sd
1      No   0.0000  465.7146  802.8571 1128.249 2391.008  803.9438 456.4762
2     Yes 652.3971 1511.6110 1789.0934 1988.870 2654.323 1747.8217 341.2668
     n missing
1 9667       0
2  333       0

Some things that were uncovered in the code above:

🔗

The probability of someone defaulting is 0.0333.
🔗

🔗
The percentage of people defaulting is 3.33%.
🔗

🔗
Note: Probabilities are always going to be a number between 0 and 1. This is incredibly important later when we are trying to create our line.
🔗

🔗
1 in every 29 people will default on their credit card. These are the odds, which by definition, compare successes (defaults) to failures (non-defaults).
🔗

🔗
While probabilities are between 0 and 1, odds are between 0 and ∞.
🔗

🔗
An overwhelming majority of students did not default.
🔗

🔗
The mean income of people who do not default is only slightly higher than people who do.
🔗

🔗
The mean balance of people who do default is more than double the amount than people who don’t.
🔗

🔗

Overall, most people don’t default on their credit cards, and while we see differences in means, we do not know if the difference is statistically significant or not.

🔗

Now that we have some insights, it is time to visualize!

🔗

Prev Top Next