This Is the Actual Point of the Project

A working launcher isn't the deliverable. Turning it into a trustworthy predictive tool is. A team that says "it works, we're done" without a real calibration curve hasn't finished the assignment.

This is the exact same workflow from Sessions 11–14 — repeated trials, structured logging, functions, conditional formatting, charting a trendline and reading R² — just with your own launcher instead of a spring, and real stakes instead of practice.

Bungee safety still applies every session. Every pull, every time, no exceptions.

Build

Step 1: Test Across a Range of Settings

Run your independent variable across a range of settings, with multiple trials at each setting. Push for enough distinct points across your intended range to actually trust interpolation — 2–3 points stretched across the whole range isn't enough.

Step 2: Log Every Trial in Sheets

Structure your table the way you learned in Session 12: labeled headers with units, one row per setting. Add Average, Min, and Max columns with real formulas, and a Spread column with conditional formatting flagging any setting where your trials disagreed too much — the same as Session 13.

Step 3: Plot and Fit a Curve

Scatter plot your IV vs. landing distance, add a trendline, display the equation and R² — direct payoff of Session 14. If R² is weak, diagnose it the same way: a measurement problem, or a real non-linearity in this range?

Step 4: State Your Usable Range

Based on your actual data, state the span of distances you're confident predicting within the ±3 yard margin. This isn't a guess — it has to be justified by what you measured.

Checkpoint

Your submitted calibration file must include all of the following:

• Enough distinct settings tested, with multiple trials each, to support real interpolation
• Every trial logged, with working Average/Min/Max/Spread formulas
• Scatter plot, trendline, and R² displayed
• A usable range that's actually justified by your data, not asserted

Reflection

Answer the following questions before submitting your work.

1. Is your R² strong or weak? If weak, is that measurement noise or a real effect — and how would you actually tell the difference at these higher stakes?
2. Did your conditional formatting flag any setting? What did you do about it?
3. Sessions 24–25 ask for a small, targeted improvement, not a redesign. Looking at your data, where is your model weakest or least precise right now?

Submit

Submit your calibration dataset, fitted curve, stated usable range, and your reflection.

Milestone Complete