When numerical data may contain entry errors, bias, or unusual values, a box-and-whisker plot gives a compact view of its distribution. It summarizes quartiles, the median, and potential extremes so you can question patterns before choosing statistical tests or models. section 2.6.1
Use it alongside a histogram rather than treating it as proof. For example, calculate the time between review creation and answer, inspect the distribution, and investigate unusually long response times against the original records. section 2.6.3
A practical inspection workflow is:
- Calculate the difference between the two timestamps and choose a useful unit, such as hours or days. section 2.6.3
- Review descriptive statistics, then compare a histogram with the box plot. section 2.6.3
- Validate apparent outliers and trends against review scores, time periods, or raw records before drawing conclusions. section 2.6.1 section 2.6.3
import matplotlib.pyplot as plt
# Histogram
df_order_reviews['time_diff'].plot(kind='hist', bins=50)
plt.xlabel('Time Difference (Seconds)')
plt.ylabel('Frequency')
plt.title('Distribution of Time Difference Between Review Creation and
Answer')
plt.show()
# Box plot
df_order_reviews['time_diff'].plot(kind='box', vert=False)
plt.xlabel('Time Difference (Seconds)')
plt.title('Box Plot of Time Difference Between Review Creation and
Answer')
plt.show()
Inspect the Python listing that creates a histogram and a horizontal box plot for timestamp differences. section 2.6.3
