Addressing Issues with Regression Assumptions
Über den Vortrag
Der Vortrag „Addressing Issues with Regression Assumptions“ von David Spade, PhD ist Bestandteil des Kurses „Statistics Part 1“. Der Vortrag ist dabei in folgende Kapitel unterteilt:
- Addressing Issues with Regression Assumptions
- Outliers, Leverage and Influential Points
- Lurking Variables and Causation
- The Logarithmic Transformation
Quiz zum Vortrag
If the regression assumptions are not violated, what do we expect to see in a scatterplot of the residuals?
- We expect to see random scatter of 0.
- We expect to see a curved pattern in the plot.
- We expect to see outliers in the plot.
- We expect to see parts of the plot where the spread is larger in some parts than it is in others.
- We expect to see small clusters of residuals.
What is an acceptable way to handle the presence of subsets or multiple groups in our data set?
- Performing one linear regression for each subgroup is the best way to handle the presence of multiple groups in our data.
- Performing one linear regression with all the data points is the best way to handle the presence of multiple groups in our data.
- Linear regression cannot be used to deal with this type of data.
- If there are multiple groups, there is no way to analyze the data set.
- Perform a logarithmic regression for the larger subgroup and a linear regression for the smaller subgroup.
If a data point has an x-value that is far away from the average x-value, what property does it possess?
- This point is said to have high leverage.
- This point is said to have high value.
- This point is said to have low leverage.
- This point is said to have low value.
- The point is said to have low volatility.
What is the best way to handle influential points?
- The best way to handle influential points is to perform two separate regressions including and excluding the influential point.
- The best way to handle influential points is to perform a single linear regression, but mention that there is an influential point present.
- The best way to handle influential points is to find another method besides linear regression to analyze the data.
- The best way to handle influential points is to discard them and perform the regression as though they never existed.
- The best way to handle influential points is to place extra emphasis on their value.
What is a common transformation to make a left-skewed unimodal distribution more symmetric?
- Squaring the response values is used to make unimodal, left-skewed distributions more symmetric.
- The log transformation is used to make unimodal, left-skewed distributions more symmetric.
- The negative reciprocal of the response values are used to make unimodal, left-skewed distributions more symmetric.
- The square root transformation is used to make unimodal, left-skewed distributions more symmetric.
- Adding a constant value to the skewed data points is the best way to make a skewed distribution more symmetric.
What is NOT a goal of transforming data?
- Make the distribution of a variable more asymmetric
- Make the distribution of a variable more symmetric
- Make the spread of several groups more alike
- Make the form of a scatter plot more nearly linear
- Make the scatter plot spread out evenly rather than thickening at one end or the other
What transformation is good for measurements that cannot be negative and for wide-ranging sets of data?
- Log transformation
- Square the response value
- Square root of response value
- Negative reciprocal
- Exponential transformation
What transformation is good for changing the direction of relationships?
- Negative reciprocal
- Square the response value
- Square root of response value
- Log transformation
- Exponential transformation
What transformation is not a data transformation technique?
- Exponential transformation
- Square the response value
- Square root of response value
- Log transformation
- Negative reciprocal
What is NOT a caution for dealing with regression assumptions?
- Make sure the relationship is quadratic
- Make sure the relationship is straight
- Do not extrapolate
- Look for unusual points, high leverage points and influential points
- Be careful if your data has multiple modes
Dozent des Vortrages Addressing Issues with Regression Assumptions
David Spade, PhD
Dr. David Spade is an Assistant Professor of Mathematical Sciences and Statistics at the University of Wisconsin-Milwaukee and holds a courtesy appointment as an Assistant Professor of Statistics at the University of Missouri-Kansas City, USA.
He obtained his MS in Statistics in 2010 and then completed his PhD in Statistics from Ohio State University in 2013.
An experienced mathemathics instructor, Dr. Spade has been teaching diverse statistics courses from the introductory to the graduate level since 2007.
Within Lecturio, he teaches courses on Statistics.
Kundenrezensionen
5,0 von 5 Sternen
| 5 Sterne |
|
5 |
| 4 Sterne |
|
0 |
| 3 Sterne |
|
0 |
| 2 Sterne |
|
0 |
| 1 Stern |
|
0 |
