AP® Statistics Unit 5 flashcardsRegression Analysis
25 cards · about 6 minutes for the whole deck
Flashcard drill
Space flips the card, arrow keys move, and keys 1 to 4 rate how well you knew it.
- Position
- 1 / 25
- Due
- 25
- Mastered
- 0 / 25
This card: New
Flip the card before you rate it.
Scatterplot
A graph of paired values of two quantitative variables, with the explanatory variable on the x-axis and the response on the y-axis.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Explanatory variable
The variable that may help explain or predict changes in the other one. It goes on the x-axis.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Response variable
The outcome variable that you measure or predict. It goes on the y-axis.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Form, direction, strength
How to describe a scatterplot: form (linear or curved), direction (positive or negative), strength (how tightly the points follow the pattern), plus unusual features.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Positive association
As x increases, y tends to increase. The points rise from left to right.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Negative association
As x increases, y tends to decrease. The points fall from left to right.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Correlation coefficient (r)
A number from −1 to 1 that measures the direction and strength of a linear relationship. It has no units.
Topic 5.2: Correlation
Properties of r
r stays the same if you switch x and y or change units. It only measures linear relationships, so a strong r doesn't prove a line is the right model.
Topic 5.2: Correlation
r near 0
No linear relationship. There could still be a strong curved relationship, so always look at the scatterplot.
Topic 5.2: Correlation
Correlation vs. causation
A strong correlation doesn't show that x causes y. A lurking variable could drive both, unless the data come from a randomized experiment.
Topic 5.2: Correlation
Regression line ŷ = a + bx
A line that predicts the response y from the explanatory variable x. b is the slope and a is the y-intercept.
Topic 5.3: Linear Regression Models
Predicted value (ŷ)
The value of y the regression line gives for a particular x. The hat means it's a prediction, not an observed value.
Topic 5.3: Linear Regression Models
Interpolation
Using the regression line to predict for an x inside the range of the data. It's usually reasonable.
Topic 5.3: Linear Regression Models
Extrapolation
Predicting for an x outside the range of the data. It's risky, because the pattern may not continue.
Topic 5.3: Linear Regression Models
Residual
Observed y minus predicted y: y − ŷ. Positive means the line underestimated, and negative means it overestimated.
Topic 5.4: Residuals
Residual plot
A scatterplot of residuals against x (or ŷ). Random scatter around 0 supports a linear model. A curved pattern means a line isn't the best fit.
Topic 5.4: Residuals
Least-squares regression line (LSRL)
The line that makes the sum of the squared residuals as small as possible. Technology finds its slope and intercept.
Topic 5.5: Least-Squares Regression
LSRL passes through (x̄, ȳ)
The least-squares line always goes through the point of means. Plug in x̄ and you get ȳ.
Topic 5.5: Least-Squares Regression
Interpreting the slope
"For each additional 1 [unit of x], the predicted [y] increases (or decreases) by b [units of y]."
Topic 5.5: Least-Squares Regression
Interpreting the y-intercept
"The predicted [y] when [x] = 0 is a." It may have no real meaning if x = 0 is impossible or far outside the data.
Topic 5.5: Least-Squares Regression
Coefficient of determination (r²)
The proportion of the variation in y that's explained by the linear relationship with x. It goes from 0 to 1.
Topic 5.5: Least-Squares Regression
Sum of residuals
For the least-squares line, the residuals always add up to 0, so the line overestimates and underestimates in balance.
Topic 5.5: Least-Squares Regression
Lurking variable
A variable not in the study that affects both x and y. It can create a strong correlation without any cause-and-effect link.
Topic 5.2: Correlation
Cluster in a scatterplot
A group of points separated from the rest. It can mean the data contain different kinds of units.
Topic 5.1: Graphical Representations Between Two Quantitative Variables
Interpreting r
"There is a [strong/moderate/weak], [positive/negative] linear relationship between [x] and [y]." Always name the variables.
Topic 5.2: Correlation