Degree vs Skills · Career Guide
Degree vs Skills: When Mathematics Background Matters—and When It Doesn't
Think of maths in data jobs as a ladder. Analysts mostly need the bottom rungs: arithmetic done carefully, percentages, averages, basic statistics. Data scientists need the middle: probability, inference and the ideas behind models. A small set of roles, mainly research and some machine learning engineering, need the top. A maths degree puts you higher on the ladder from day one. Whether that matters depends on which rung your job sits on.
Data analyst: careful basics beat advanced theory
Most analyst work uses school-level maths applied carefully, plus introductory statistics: mean, median, percentiles, variance, correlation, and knowing when a difference might be noise. The errors that cause trouble are rarely about advanced maths. A typical one is averaging percentages:
| Store | Visitors | Buyers | Conversion |
|---|---|---|---|
| Store A | 1,000 | 100 | 10% |
| Store B | 100 | 30 | 30% |
| Combined | 1,100 | 130 | 11.8% |
Averaging the two rates gives 20%, which badly overstates the real combined rate of 130 / 1,100 = 11.8%. An analyst who catches this is using maths well, degree or not. Verdict: a maths degree helps a little; a careful habit with numbers and a basic statistics course cover what most analyst jobs need.
Data scientist: statistics is the job
Data scientists design experiments, build models and judge whether results are real. That needs probability, hypothesis testing, regression, and the intuition behind algorithms (why regularisation helps, what a loss function is). Here is the kind of calculation that comes up when reading an A/B test:
from math import sqrt, erf
n1, c1 = 2000, 100 # control: visitors, conversions
n2, c2 = 2000, 130 # variant
p1, p2 = c1/n1, c2/n2
p = (c1 + c2) / (n1 + n2) # pooled rate
se = sqrt(p * (1 - p) * (1/n1 + 1/n2))
z = (p2 - p1) / se
p_value = 2 * (1 - 0.5 * (1 + erf(abs(z) / sqrt(2))))
print(f"control={p1:.3f} variant={p2:.3f} z={z:.2f} p={p_value:.3f}")
control=0.050 variant=0.065 z=2.04 p=0.042
Knowing the code is the easy part. A data scientist should also be able to say what the p-value does and does not mean, whether the test was planned with enough sample size, and why checking results every day and stopping when p dips below 0.05 inflates false positives. Verdict: a maths or statistics degree is a genuine advantage here. Without one, plan months of serious statistics study, not a weekend.
AI engineer: applied maths, learned as needed
Many AI engineer roles today build applications on top of large language models. The maths is applied: vectors and cosine similarity for embeddings, probability for sampling settings like temperature, and evaluation metrics such as precision and recall. Training or modifying models needs more: linear algebra, calculus for gradients, optimisation. Verdict: for application-focused roles, a maths degree is helpful but not expected; for research roles it often is. How much maths AI engineers really need goes through this in detail.
Summary by role
| Role | Maths used most | How much a maths degree helps | If you lack one |
|---|---|---|---|
| Data analyst | Arithmetic, percentages, descriptive statistics | A little | An introductory statistics course and practice with real data |
| Data scientist | Probability, inference, regression, model intuition | A lot | Structured study of statistics and probability over several months |
| AI engineer (applications) | Vectors, similarity, evaluation metrics | Some | Learn the specific topics alongside projects |
| ML or research scientist | Linear algebra, calculus, optimisation, statistics | Often required | A formal degree or equivalent depth is usually expected |
If you do not have a maths background
- Start with the maths of your target role, not all of maths. For analysts that is descriptive statistics and probability basics.
- Learn each idea through a dataset: compute a median, a percentile, a correlation, and explain what it tells you.
- For data science, work through a statistics textbook with exercises, then a statistical learning book.
- Use maths in your projects visibly: confidence intervals in a report, a correctly analysed A/B test, a model evaluated with the right metric.
Arts and commerce graduates make this transition regularly; see whether an arts graduate can become a data scientist. For how the three roles differ beyond maths, read AI engineer vs ML engineer vs data scientist, and if you are considering formal study, whether a master's helps in data science.