### Objective

The objective of this activity is to use data from the 2021 Speed Climbing and Lead Climbing Olympic to investigate relationships, create regression models, make predictions, and examine residuals.

During the 2021 Summer Olympics in Tokyo, competition climbing became an Olympic sport for the first time ever. Athletes competed in three separate disciplines: speed climbing, bouldering, and lead climbing. 20 men and 20 women competed in the 2021 Olympic Sports Climbing Qualification Rounds.

In speed climbing, two climbers race side-by-side to scale identical routes on a 15m high wall set at an angle of 95 degrees.

The walls used for bouldering present a range of challenges, with overhangs and some holds so small that they can only be held by the fingertips. Climbers must plan each move carefully while constantly being aware of the 4 minute time limit. The goal is to complete as many routes as possible.

When lead climbing, athletes wearing harnesses attached to a climbing rope attempt to climb as high as they can on a taller wall measuring 15-20 meter within six minutes. The wall features 40-60 handholds. Climbers are scored on how far they progress, with each handhold earning 1 point.

The table below data gives about the three disciplines for each athlete in the women's qualification rounds. The data include the time, in seconds, of the speed climb, the number of holds attained in lead climbing, and their ranking in the bouldering competition.

1. What do you notice?
2. Student answers will vary. Encourage students to make observations that relate two variables. Interesting observations: The athlete with the fastest climb in speed climbing had the lowest rank in bouldering. The athlete with the slowest climb in speed climbing attained the fewest holds in lead climbing. Several of the highest ranked bouldering athletes were in the bottom half of the speed climbing times.

3. What do you wonder?
4. Student answers will vary. "I wonder if an athlete that is good at speed climbing is not as good at bouldering?" "I wonder if athletes will perform similarly in speed climbing and lead climbing?"

Athlete Speed (s) Lead (holds) Bouldering (rank)
MIROSLAW Aleksandra 6.97 12 20
JAUBERT Anouck 7.12 16 13
SONG Yiling 7.46 13 19
NONAKA Miho 7.55 30 8
KAPLINA Iuliia 7.65 14 18
YIP Alannah 7.99 21 16
CONDIE Kyra 8.08 22 11
CHANOURDIE Julia 8.17 25 15
NOGUCHI Akiyo 8.23 27 3
KLINGLER Petra 8.42 16 10
PILZ Jessica 8.51 33 9
MACKENZIE Oceania 8.83 15 12
GARNBRET Janja 9.44 26 1
MESHKOVA Viktoriia 9.54 29 6
COXSEY Shauna 9.65 21 4
SEO Chaehyun 10.01 40 5
KRAMPL Mia 10.43 26 14
ROGORA Laura 10.5 25 7
STERKENBURG Erin 11.1 7 17

Source: Olympics.com

### Making Predicitons

We wish to investigate the relationship between the athlete's time in Speed Climbing and the number of holds they attained during Lead Climbing. How would you predict the relationship to behave? Would you expect a fast climber in Speed Climbing to attain a high number of holds or a low number of holds in Lead Climbing? Explain your reasoning.

Answers will vary. "Since both of these disciplines require climbing high walls, I predict that athletes with short climb times in Speed Climbing will attain higher holds in Lead Climbing."

### Scatterplot and Correlation

1. Use technology to make a scatter plot of the data on time and number of holds and describe the association between the two variables.
2. A scatterplot can be constructed using the Regression application. Input the time in X1 and the number of holds attained in Y1. Use the Graph tab to view the scatterplot.

There appears to be a weak, positive, linear relationship between the time in Speed Climbing and the number of holds in Lead Climbing. A possible outlier is the athlete with a time of 11.1 seconds and 7 holds.

3. Determine the correlation coefficient for the data and describe what it means in regards to the data.
4. The correlation coefficient is reported in the legend of the scatterplot, r = 0.2294. This confirms what we saw in the scatterplot, there is a weak, positive, linear relationship between speed climbing time and holds attained in lead climbing.

5. Fit a linear function to model the relationship of speed and hold attained. Write the equation of your model and interpret the slope in context.
6. The linear model is reported in the legend of the scatterplot. The model is $\hat{y}=1.546x + 8.614$ where $\hat{y}$ is the predicted number of holds attained in lead climbing and $x$ is the time in seconds in speed climbing. A slope of $1.546$ means that for every additional one second the climber took to copmlete the speed climbing event, they are expected to obtain 1.546 additional holds on average.

### Regression and Residuals

To investigate the relationship between the athlete's time in Speed Climbing and their rank in Bouldering, a scatterplot and linear model was created as shown below.

The linear function

$R(t) = -1.75t + 25.753$

has been suggested as a good fit for the data. The table below displays the 20 athlete's time in seconds during Speed Climbing.

1. Complete the table by computing the predicted ranks for each athlete based on the model. Then compute the residual.
2. Time (s) Rank Predicted Rank Residual
6.97 20
7.12 13
7.46 19
7.55 8
7.65 18
7.99 16
8.08 11
8.17 15
8.23 3
8.42 10
Time (s) Rank Predicted Rank Residual
8.51 9
8.67 2
8.83 12
9.44 1
9.54 6
9.65 4
10.01 5
10.43 14
10.5 7
11.1 17

The Predicted Rank is computed using the model $R(t) = -1.75t + 25.753$. For example, the predicted rank for the athlete whose time was 6.97 is $-1.75(6.97) + 25.753 = 13.556$. A formula can quickly compute all predicted values. In the Regression app, fill X2 and Y2 using the formulas $X2 = X1$ and $Y2 = -1.75(X2) + 25.753$

The Residuals are computed using Rank - Predicted Rank. Replace Y1 with the athlete's Rank. Then fill X3 and Y3 using the formulas $X3 = X1$ and $Y3 = Y1 - Y2$

Time (s) Rank Predicted Rank Residual
6.97 20 13.556 6.445
7.12 13 13.293 -0.293
7.46 19 12.698 6.302
7.55 8 12.541 -4.541
7.65 18 12.366 5.635
7.99 16 11.771 4.230
8.08 11 11.613 -0.613
8.17 15 11.456 3.545
8.23 3 11.351 -8.351
8.42 10 11.018 -1.018
Time (s) Rank Predicted Rank Residual
8.51 9 10.861 -1.861
8.67 2 10.581 -8.581
8.83 12 10.301 1.700
9.44 1 9.233 -8.233
9.54 6 9.058 -3.058
9.65 4 8.866 -4.866
10.01 5 8.236 -3.236
10.43 14 7.501 6.500
10.5 7 7.378 -0.378
11.1 17 6.328 10.672
3. Interpret the value of the residual for the athlete with a time of 8.23 seconds and ranked 3rd in bouldering.
4. The residual for the athlete with a time of 8.23 seconds and ranked 3rd in bouldering is $11.351 - 3 = -8.351.$ This means that this athlete ranked about 8 positions lower (better) than expected by the model, based on their time.

5. For which athlete did the model under predict by the most?
6. The model "underpredicts" when the predicted value is less than the observed value. Therefore, we are looking for the largest, positive residual. This occurred for the athlete with a time of 11.1 seconds and ranked 17. The model predicted an athlete with a time of 11.1 seconds would rank around 6th. Thus the model underpredicted by about 11 positions.

7. Create a residual plot for the linear regression of rank on speed for the 20 athletes.
8. Clear the X1 and X2 columns leaving only X3, Y3. Use the Graph tab to view the Residual Plot

9. Use the residual plot to determine the goodness of fit of the linear function for the data.
10. There appears to be a pattern in the residual plot. Therefore, a non-linear model is more appropriate.

### A Better Model

Perform a quadratic regression on the athlete's time in Speed Climbing and their rank in Bouldering. Assess the fit of the quadratic function by plotting and analyzing the residuals.

The regression model can be changed to quadratic by selecting Regression in the Graph tab and choosing Quadratic for the model type. The residual plot using a quadratic regression appears to be more randomly scattered than the linear model. This suggests the quadratic model is more appropriate.