Bar & Pie Charts – Advanced Quantitative Aptitude
Audience – Undergraduate students (MIT‑level) or competitive‑exam aspirants who already know the basics but want a deeper, proof‑based understanding.
Goal – Equip you with rigorous principles, edge‑case handling, and a challenging problem that blends theory with real‑world application.
1. Foundations & Definitions
| Concept | Formal Definition | Key Takeaway |
|---|---|---|
| Data Set | A finite multiset (D={d_1,d_2,\dots,d_n}) of real numbers (or categorical labels). | All visualisations start from an underlying data set. |
| Frequency | For a categorical value (c), (f(c)= | {d_i\in D: d_i=c} |
| Relative Frequency | (r(c)=\frac{f(c)}{ | D |
| Bar Chart | A collection of rectangles (bars) with base aligned to a common axis. Height (h_i) of bar (i) is a function of the associated value (v_i). | The visual area is proportional to (v_i). |
| Pie (Doughnut) Chart | A circle partitioned into sectors whose central angles (\theta_i) satisfy (\theta_i\propto v_i). | The area of a sector is (\frac{\theta_i}{2\pi}\cdot\pi r^2=\frac{\theta_i}{2}r^2); thus the sector area is also proportional to (v_i). |
| Stacked Bar | Bars partitioned into sub‑bars that sum to the total height. | Useful for multivariate categorical data. |
| Grouped Bar | Bars for multiple categories laid side‑by‑side for the same group. | Enables direct comparison. |
2. Theoretical Underpinnings
2.1 Bar Height & Area
For a bar of width (w) and height (h), the area is (A = w\cdot h).
In a bar chart we often normalise (w) to a constant (e.g., (w=1)) so that (A = h).
Proof of Proportionality
If two bars have values (v_1, v_2), we map them via a linear scaling function (S(v)=k\cdot v).
The scaling factor (k) is chosen such that the tallest bar fits the vertical axis (e.g., (k=\frac{H_{\max}}{v_{\max}})).
Then (h_i=S(v_i)=k v_i) → (h_i/h_j = v_i/v_j).
Edge Cases
- Zero / Missing Data – Set (h=0) or omit the bar.
- Negative values – Draw bars below the baseline; interpret as “deficit”.
- Large ranges – Use a logarithmic scale to avoid “crowding”.
2.2 Pie Slice Angles
Let the total value be (V=\sum_i v_i).
The slice angle (\theta_i) is defined by
[ \theta_i = \frac{v_i}{V}\cdot 360^\circ ]
or, in radians,
[ \theta_i = \frac{v_i}{V}\cdot 2\pi. ]
Proof
The sum of all slice angles is (360^\circ) (or (2\pi) rad), since
[ \sum_i \theta_i = \frac{1}{V}\sum_i v_i \cdot 360^\circ = 360^\circ. ]
Thus the mapping is linear and preserves proportions.
2.3 From Data to Chart – The Transformation Matrix
Define a transformation matrix (T) that maps a vector of raw values ( \mathbf{v} = (v_1,\dots,v_n)^T) to a vector of visual attributes:
| Chart | Attribute | Formula |
|---|---|---|
| Bar | Height (h_i) | (h_i = k h , v_i) |
| Pie | Angle (\theta_i) | (\theta_i = 360^\circ,\frac{v_i}{\sum_j v_j}) |
Here (k h) is a scaling constant that ensures the chart fits in the allotted space.
The transformation is linear in (v_i), so it can be represented as
[ T: \mathbf{v} \mapsto \mathbf{a}, \qquad