Lesson 21: Bootstrap Confidence Intervals

April 9, 2018

Review:

• Seaborn scatter plots with “lmplot”
• Schedule for this week and next
  • Mon Apr 9 (today) – Bootstrap Confidence Intervals
  • Wed Apr 11 – Pairs bootstrap for linear regression and review for Exam 2 Redo
  • Mon Apr 16 – Exam 2 Redo
  • Wed Apr 18 – Term Project Assignment

Presentation:

• What is bootstrapping?
  • Random sampling from the original data set with replacement
• Generating bootstrap samples and replicates
  • np.random.choice()
• Calculating bootstrap confidence intervals
  • repeated sampling and statistical calculation with “for loops”
  • use np.percentile() to calcuate confidence intervals
• Example: https://repl.it/@justinholman/SellingPriceBootstrapConfidenceIntervals

Activity:

• Write a program in repl.it to generate bootstrap confidence intervals of “Selling Price per SqFt” (PPSF)
  • Use your real estate data
  • Use np.random.choice() to repeatedly sample 100 properties, repeat 1,000 times
  • Calculate the mean PPSF (i.e., divide “Selling Price” by “Total SqFt”); again, repeat 1,000 times
  • Save each mean PPSF in a list (this is just like the dice game…remember?)
  • Produce a histogram of the resulting mean PPSF
  • Calculate a 95% confidence interval of mean PPSF (i.e., find the 2.5 percentile and the 97.5 percentile)

Assignment:

• Complete the Statistical Thinking in Python (Part 2) course.