Modelling with Straight Lines

• The Meaning of Gradient ($m$): In a real-world context, the gradient represents the rate of change of the dependent variable per unit increase of the independent variable. For example, if $x$ is time in hours and $y$ is distance, $m$ represents velocity in units per hour.

• The Meaning of Intercept ($c$): The y-intercept represents the initial value or the value of $y$ when $x = 0$. This often corresponds to a fixed cost, a starting position, or a baseline measurement before a process begins.

• Linear Homogeneity: Some models pass through the origin ($c = 0$), implying direct proportionality. However, many practical models include a non-zero $c$ to account for setup conditions or constant overheads.

• Predictive Power: By substituting values for $x$, we can interpolate to find missing values within the known data range or extrapolate to predict future outcomes beyond the measured data.

Step-by-Step Modelling Process

• Identify Variables: Determine which physical quantities are changing and assign them to $x$ (independent) and $y$ (dependent) axes based on which variable drives the change.
• Find the Constants: Use provided data points to calculate the gradient $m = \frac{y_2 - y_1}{x_2 - x_1}$ and substitute a point into $y - y_1 = m(x - x_1)$ to find the full equation.
• Interpret Context: Always map the numerical values of $m$ and $c$ back to the specific units of the problem to ensure the mathematical result makes physical sense.
• Solve for Unknowns: Set $y$ to a target value to find the required $x$, or vice versa, ensuring that algebraic manipulation follows the standard rules of linear equations.

• Discrete vs. Continuous Data: While the line itself is continuous, the real-world application might only be valid for whole numbers (discrete), such as number of people or items sold.

• Interpolation vs. Extrapolation: Predicting values within your data set is generally reliable, whereas predicting far into the future (extrapolation) is risky as the linear relationship may eventually break down.

• Check the Units: Examiners often provide data in one unit (e.g., minutes) but ask for a rate in another (e.g., hours); always convert units before calculating the gradient to avoid scale errors.

• State Your Variables: At the start of a modelling question, explicitly define what $x$ and $y$ represent, including their units, to clarify your logic for the marker and yourself.

• Sanity Check Constants: If a model for temperature suggests an initial value of $1000^{\circ}C$ for a cup of tea, you have likely swapped your $x$ and $y$ coordinates or miscalculated the intercept.

• Interpret the 'c': If an exam asks to 'state the meaning of the constant $c$', do not just say 'the y-intercept'; instead, describe it as the 'fixed fee before any usage' or 'the starting height'.

• Assuming Infinite Linearity: Real-world processes often have limits (e.g., a spring will eventually break), so a linear model is usually only valid within a specific domain of $x$ values.

• Swapping Variables: Incorrectly identifying which variable is independent can lead to a gradient that is the reciprocal of the correct answer, completely changing the model's interpretation.

• Misinterpreting Zero Intercepts: Students often assume a model must pass through $(0,0)$; however, most real-world scenarios have a starting cost or value that must be accounted for by a non-zero $c$.

• Physics Applications: Linear models are foundational in kinematics (displacement-time graphs where gradient is velocity) and Hooke's Law (force-extension where gradient is the spring constant).

• Economic Analysis: In business, these models separate fixed costs from variable costs, allowing for break-even analysis and profit margin projections.

• Statistical Foundations: This topic is the prerequisite for Linear Regression, where mathematicians find the 'line of best fit' for noisy data that isn't perfectly linear.