Statistics for Data Science and Business Analysis - A4. No Autocorrelation

The video tutorial discusses the concept of serial correlation, particularly in time series data like stock prices. It highlights the day of the week effect, where stock returns vary between Fridays and Mondays, and explores explanations by Merton Miller and Kenneth French. The tutorial explains that linear regression assumes no autocorrelation, which is often not the case in time series data. It suggests plotting residuals to detect patterns and recommends using alternative models like autoregressive or moving average models when autocorrelation is present.

What is a common phenomenon observed in stock prices related to serial correlation?

• The year-end effect

• The holiday effect

• The month-end effect

• The day of the week effect

Why might errors on Mondays be biased downwards in stock price regressions?

• Markets react to delayed bad news from the weekend

• Investors are more optimistic on Mondays

• Mondays are generally more volatile

• Firms release good news on Mondays

What assumption does linear regression make about error distribution?

• Errors follow a normal distribution

• Errors are randomly spread around the regression line

• Errors are correlated with time

• Errors are always positive

Which model is NOT suitable for handling autocorrelated error terms in time series data?

• Autoregressive integrated moving average model

• Moving average model

• Autoregressive model

• Linear regression model

What is a common method to detect autocorrelation in residuals?

• Calculating the mean of residuals

• Plotting residuals on a graph to look for patterns

• Performing a t-test on residuals

• Using a histogram of residuals