IB Mathematics Applications & Interpretation (AI) HL Internal Assessment Example

The introduction is strongly rooted in a genuine personal context, which gives the exploration a clear purpose from the outset. Because the motivation is specific and first-hand, the mathematical modelling feels meaningful rather than generic.

This conclusion overstates what the fit comparison can show. Better $R^2$, lower RMSE, and cleaner residuals support a better descriptive model, but they do not prove the underlying physiology. The student should replace “prove” with more cautious language such as “suggests” or “is consistent with.”

The student could strengthen the methodology by justifying the sampling interval more rigorously. Explaining why 5-second measurements are appropriate for heart-rate recovery, and how this interval balances precision against measurement noise, would make the method feel more deliberate.

This is a strong causal claim that is not yet justified by the evidence. The dataset is too small to establish a direct fitness effect, and other variables such as age and individual recovery patterns are confounded with activity level. The student should frame this as an observed pattern rather than a causal conclusion.

What else, besides training status, could explain the ordering of the fitted $\tau$ values? The student should consider whether age, baseline heart rate, or repeated-session variation might also influence the pattern before claiming that the result “proves” the assumption.

Calling the model “perfect” is inconsistent with the very residual pattern described in the same sentence. The student should emphasise that high $R^2$ alone is not enough, and that the residuals reveal limitations in the single-exponential model.

This is a good forward-looking reflection, but it would be stronger if the student named one concrete next step. For example, they could propose testing the model on a larger sample, comparing a different decay model, or examining uncertainty in $\tau$ values.

This definition is inaccurate because $HR$ is the heart rate at time $t$, not the initial heart rate. The student should define the initial value separately, for example as $HR(0)$, to keep the notation mathematically correct.

This is a strong mathematical setup because the assumption is translated clearly into an ordinary differential equation. The student shows the link between the verbal idea of proportional decay and the algebraic model, which makes the reasoning easy to follow.

The renaming step is too imprecise to support the derivation cleanly. The student should show exactly how the integration constant is absorbed into the amplitude term and explain why $k=\frac{1}{\tau }$ is introduced, rather than presenting it as a simple label change.

The phrase “transparent fitting process” is too absolute, and the sentence needs cleaner mathematical wording. The student should avoid implying that software output is automatically fully transparent; instead, they could say that Desmos provides a convenient way to perform nonlinear regression on the same data.

The worked example should be checked for internal consistency, because the surrounding calculation later contains a residual value that appears to change from the value stated earlier. The student should verify every intermediate result so the calculation chain remains reliable.

The student could make this comparison more mathematically convincing by adding a complexity-aware measure. An adjusted $R^2$, AIC, or a nested-model test would help justify whether the extra bi-exponential parameters are warranted.

Which version of the two-sample $t$-test is being used here, and are its assumptions satisfied by the three repeated sessions per participant? The student should clarify whether the test is pooled or Welch’s, and justify the degrees of freedom used later in the analysis.

This statement is too definitive given the small repeated dataset and the uncertain test design. The student should report the result more cautiously and explain what statistical conclusion is justified by the chosen test rather than presenting significance as final proof.

This application of the $t$-test is not well justified. The three sessions for each participant are repeated measurements, so they may not be independent samples. The student should address dependence and consider a repeated-measures or ANOVA-based approach instead.

The practical discussion would be stronger if the student connected $au$ more explicitly to decision-making. For instance, they could explain how changes in $au$ might inform training load, recovery timing, or monitoring progress over several weeks.

This overstates the interpretation of the fitted terms. The model may match the data better, but the parameters do not automatically separate physiological mechanisms in a direct or unique way. The student should distinguish modelling convenience from biological proof.

The student shows clear independent thinking by moving from a source study to a personally collected modelling task. This is more than imitation: the investigation is adapted to the student’s own training context and data, which gives the work authentic ownership.

This demonstrates initiative because the student first tests the fitting workflow on published data before applying it to the personal investigation. That sequence shows planning and a willingness to validate the method before relying on it.

This is a strong sign of independent analysis because the student does not stop at a high $R^2$. Instead, they use the residual pattern to question the simpler model and revise the approach, which is exactly the kind of initiative that strengthens an exploration.

The student’s interpretation is creative, but here it becomes too confident for the evidence provided. The observed fit differences do not by themselves establish that fitter individuals physiologically recover in two distinct phases; that claim needs either stronger evidence or softer wording.

The student could show more initiative by proposing a specific extension of the investigation. A follow-up comparison across a larger sample, a different recovery model, or an uncertainty analysis would make the ending feel more original and purposeful.

This is a strong reflective moment because the student notices a limitation in the first model rather than accepting a high fit statistic blindly. Recognising the S-shape shows meaningful critical awareness of what the residuals reveal about model structure.

The reflection here becomes too conclusive too quickly. The student should reflect on whether improved fit alone justifies a physiological claim, and whether alternative explanations such as overfitting or sample-specific behaviour could also account for the result.

The student is beginning to connect the fitted parameter to broader interpretation, which is valuable reflection. Comparing personal results to the published study helps situate the investigation and shows that the student is thinking beyond a single data set.

This goes beyond what the data can support. The student should reflect more cautiously on what the fitted $\tau$ can actually indicate and avoid turning a descriptive comparison into a broad physiological generalisation.

What would change in the interpretation if the three sessions per participant are not independent observations? The student should reflect on whether the test still supports the same conclusion once repeated-measures dependence is considered.

This is not justified by the investigation. The data compare recovery curves, not mortality outcomes, so the student should separate the published association from what their own results can actually conclude.

The student should reflect on the limits of inference from a three-person sample. The ordering is interesting, but it does not prove a general relationship between age, fitness, and recovery speed; it only describes this small set of participants.

The limitations section would be stronger if each limitation were paired with a consequence for interpretation. The student could explain how each issue affects confidence in the conclusion, not just list the limitation itself.

The mathematics is relevant and appropriately ambitious for HL because the student starts from a first-order differential equation and builds a modelling framework from it. This is a sound way to connect theory to the real data set.

This step contains a notation error in the resting-rate term, which obscures the mathematics. The student should correct the variable substitution carefully so that the derivation is valid and easy to follow.

Using technology is fine, but the student should still show some mathematical understanding of what the regression is optimising. A brief explanation of nonlinear least squares would strengthen the mathematical depth and HL rigour.

The student could strengthen the model comparison by explaining why the extra parameters are worth it mathematically. A criterion such as adjusted $R^2$, AIC, or a formal nested-model comparison would better address model complexity.

This conclusion is mathematically too strong because better in-sample fit does not establish model superiority once complexity is considered. The student should distinguish empirical fit from model validity and avoid using descriptive statistics as proof.

The mathematics would be stronger if the student explained how repeated sessions were handled statistically. Treating nine fits as nine independent observations without discussing dependence weakens the validity of later inference.

The residual interpretation is correct in direction, but the student should state the sign convention consistently and avoid mixing symbolic and verbal notation imprecisely. Small notation care like this matters because it affects the reliability of later calculations.

The hypothesis-test setup is not rigorous enough for the data structure. Because the measurements come from repeated sessions on the same participants, the student should justify independence or use a method designed for repeated measures.

The conclusion is mathematically overconfident given the small sample and the test assumptions. The student should report the result as evidence within a limited model, not as a definitive statistical proof.

A stronger HL treatment would connect the fitted parameter to uncertainty. The student could discuss confidence intervals or parameter error estimates to show how precisely $\tau$ is known, not just what it means.

The parameter interpretation is presented as established fact, but the fitted terms do not uniquely identify physiological processes without further justification. The student should acknowledge that the model is descriptive and that physiological interpretation is an inference, not a direct measurement.