Qs D

• Correlation coefficient


  A superior metric than covariance to measure the linear relationship, as it is not affected by the magnitude of values in data
• Probability distributions


  • Weibull
  • Log-normal
  • Logistic
  • Normal
• Calculating probability of not having congestion at a gate

  1. Calculate λ (arrival rate)

  2. Use exponential distribution to calculate P(x>8)
• Calculating probability of conducting 20 experiments before identifying no pollutants 18 times

  Use negative binomial distribution
• Increase in advertising by a firm

  Demand curve shifts right, increasing equilibrium price and quantity
• Marginal utility added up for each unit gives total utility
• Correlation coefficient


  Informs about causation, unlike covariance
• Correlation coefficient is not affected by outliers, unlike covariance
• Covariance is difficult to calculate, while correlation coefficient is more practical
• Correlation coefficient is a superior metric than covariance by definition
• Exponential distribution is used to model the arrival of pedestrians
• Negative binomial distribution is used to calculate the probability of conducting 20 experiments before identifying no pollutants 18 times
• Calculating the upper limit of 95% confidence interval for the number of traffic congestions per day


  Use the formula: X̄ + Zα/2 * σ/√n
• Calculating the hypothesis test score for the mean number of accidents per 4 months

  Use the Z-score formula: (X̄ - μ) / √μ
• The null hypothesis is that the sample mean is equal to the population mean, and the alternative hypothesis is that the sample mean is different than the population mean
• The test score calculated is 1.83
• If the test score is 1.83

  The null hypothesis cannot be rejected at 90% CL
• Calculating the test statistic (χ2) to test if Amsterdam and Enschede have the same crash patterns

  Use the formula: Σ (Oi - Ei)2/Ei
• The calculated χ2 is 36.37, which is greater than χ0.10,3^2 = 6.251
• Based on 90% CL, the number of crashes in Amsterdam and Enschede are different
• Conditions where Poisson regression model is preferred

  • If the dependent variable takes count values
  • If there is an overdispersion problem
• Speed limit variable has a positive coefficient and traffic volume variable has a negative coefficient in a negative binomial regression model for cycling accidents


  The higher the speed limits, the larger the number of accidents
• Low speed limit roadways do not necessarily have large traffic volume