ScienceXpertseTut.
GRADE 12 MATHEMATICS
STATISTICS
CHECKLIST Scatter plots.
Regression line and equation.
Correlation coefficient.
Outliers
BIVARIATE DATA: SCATTERPLOT
In grade 11, univariate data was used e.g. comparing ONE variable across a range of students/people/groups.
The data was obtained for Physical Sciences leaners’ marks in the Grade 12 class.
In grade 12, we start looking at bivariate data, that is, data that compares two variables within a group.
For example, comparing leaners’ Physical Sciences and Mathematics Trial Exam Marks.
NB: The two variables related to the same person, same group.
We use scatter plots in order to compare bivariate data.
SKETCHING A SCATTERPLOT
• For one to sketch a scatter plot, bivariate data is given and it is important to note that, the FIRST ROW of data
is taken to the x-axis and the SECOND ROW of data is taken to the y-axis. STANDARD PROCEDURE EVEN
WHEN NOT TOLD.
• ONE dot on a scatter plot represents ONE leaner/student/person/group and the ONE dot compares two
variables by using BOTH THE x-axis and the y-axis. We will see this shortly….
• If data is given in column format (rarely), first coloumn (x) and second column(y)
• We DO NOT have INDEPENDENT AND DEPENDENT VARIABLES in SCATTERPLOT STATISTICS.
• We will explain why…let us look at an example…….what is curve fitting ?
SKETCHING A SCATTERPLOT
SKETCHING A SCATTERPLOT
NOTES TO REMEMBER:
1. ONE dot must represent BOTH the x and y-values across the SAME person/object e.g. the car
above. ONE car has a data value on the x and y axis and is represented by ONE point.
2. After sketching the dots, the least squares regression line must be drawn. REMEMBER, it must
pass through the MEAN point and the y-intercept for an accurate estimate.
3. DRAW the line in conjunction with the first point and last point…
4. CHECK if you have NOT extrapolated BY mistake.
DETERMINING THE REGRESSION
EQUATION
• Also known as the least squares regression line
• The regression line in grade 12 follows a linear pattern hence we draw a straight line.
• Equation looks like: y= mx + c BUT it is not, works the same though…..
• Regression equation is y-hat = a + bx where a=y-intercept, b=gradient
• NB: write the full value as per the calculator before rounding off….
THE REGRESSION EQUATION
• It is important to note that the regression equation ESTIMATES the relationship and how the x-
variable MAY influence the y-variable….
• Correlation DOES NOT equal causation…..example ?
• We can estimate a y-value given an x-value and vice versa…NOTË: These are ESTIMATE which we
can check if they are reliable or not….will speak about that soon….
• Estimating a value within our data set is known as INTERPOLATION
• Estimating a value outside our data set is known as EXTRAPOLATION
• NOT ACCURATE REPRESENTATION JUST ESTIMATES…
• Which ONE do you think is more inaccurate or prone to giving us a completely wrong estimate ?
INTERPOLATION AND EXTRAPOLATION
1. Estimate the age of a new car that has just been brought today.
2. Estimate the age of a 15 year-old car. Are the two examples INTERPOLATION or EXTRAPOLATION
3. How old is a car costing R250 000 ? What is the price of a car that is 5-years old in this dealership ?
4. Is the previous question showing an example of INTERPOLATION or EXTRAPOLATION ?
INTERPOLATION AND EXTRAPOLATION
• We can also interpolate/extrapolate using the SCATTER PLOT itself rather than the regression
equation.
• Still, they remain as ESTIMATES not accurate measures….do you remember why ?
• We use dotted lines to show extrapolation of the least squares regression line. It is very important
to note this.
• Like we did in ogives, match the y-value with the corresponding x-value and vice versa.
• If asked to extrapolate/interpolate using the scatter plot, PLEASE SHOW THIS ON THE SCATTER,
that has marks [method mark].
THE STRENGTH OF A LINEAR
RELATIONSHIP
• We can use our own eyes to see whether a linear relationship is strong or weak. For you to
understand this, do you still remember what a linear relationship is….?
• The strength of a linear relationship basically examines the possibility of the scatters/dots forming
a straight lines (linear relationship). In other words, how likely it is that all points can be joined with
one straight line……
• We can observe the pattern with our eyes but up until a certain point…hence we need a way of
being consistent ……someone could see strong but someone weak in some cases…
THE
STRENGTH OF
A LINEAR
RELATIONSHIP
THE
STRENGTH OF
A LINEAR
RELATIONSHIP
THE
STRENGTH OF
A LINEAR
RELATIONSHIP
THE
STRENGTH OF
A LINEAR
RELATIONSHIP
DESCRIBING A LINEAR RELATIONSHIP
• Increase in x leads to an increase in y….POSITIVE LINEAR RELATIONSHIP
• Increase in x leads to decrease in y….NEGATIVE LINEAR RELATIONSHIP
THE CORRELATION COEFFICIENT
• Can be calculated using YOUR CALCULATOR.
• It describes BOTH the relationship and its STRENGTH
• We use small letters (r) to indicate the correlation coefficient.
• The SIGN of the correlation coefficient (r) explains/described the relationship.
• If r = positive number then there is a POSITIVE LINEAR RELATIONSHIP…remember what this means?
• If r=negative number then there is NEGATIVE RELATIONSHIP…remember what this means….?
• The number indicates the STRENGTH…see next page….
THE CORRELATION COEFFICIENT
THE CORRELATION COEFFICIENT
• PLEASE REMEMBER TO WRITE THE CORRELATION COEFFICIENT FULLY WITH ALL THE DECIMAL DIGITS.
• A stronger correlation coefficient makes INTERPOLATION reliable…while a weaker one suggests estimate is
not reliable.
• Extrapolation is generally not reliable so we cannot make a reliable estimate even with strong correlation
unless we know or are told the pattern continues as is….
OUTLIERS
• Do you still remember what an outlier is….?
• You would not be expected to calculate an outlier really here..but know how to spot it ?
• What do you this is the effect of the outlier on the regression equation and its strength ?
• What do you think happens when the outlier is below compared to when it is above the regression
line ?
FINAL WORDS
YOU SHOULD BE ABLE TO :
• Sketch a scatter plot.
• Determine the equation of the least squares regression line.
• Sketch the least squares regression line on your scatter plot.
• Use the equation of the least squares regression line OR the scatter plot to
interpolate or extrapolate.
• Calculate the correlation coefficient and INTERPRET it.
• Use the correlation coefficient to comment on the reliability of estimates.
• Understand the effect of an outlier on the least squares regression line.
QUESTIONS