WhatsApp us

College and engineering probability & statistics, online

Probability and statistics tutor online: Ajay Vatsyayan

Ajay Vatsyayan teaches calculus-based probability and statistics one to one online for college, university and engineering students. Sessions run from random variables and distributions through joint distributions and the central limit theorem to estimation, confidence intervals, hypothesis tests and the basics of regression, with each idea tied to a question you could meet in your own exam.

Random variablesDistributionsJoint distributionsCentral limit theoremConfidence intervalsHypothesis testsRegression
Ajay Vatsyayan, senior maths tutor at Maths Bodhi, Gurugram

1:1

Online sessions

Theory + method

Why a test works, not just which button

Your syllabus

Matched to your course and notation

I could do school probability with counting and tree diagrams, but now everything is a density function and an integral, and I'm lost.

I can follow the steps of a t-test, but I couldn't tell you what a p-value actually means or which test my exam question wants.

My engineering paper has a full probability and statistics unit, and joint distributions and the normal approximation keep costing me marks.

Which course

Probability, mathematical statistics or an engineering paper: know which one you are taking

'Probability and statistics' covers several different courses. A first probability course is built on calculus: random variables, densities, expectation and the standard distributions, ending with limit theorems. A mathematical statistics course then uses that machinery for estimation and inference. An engineering course compresses both into one semester, with more weight on using distributions and tests than on proofs. Ajay checks which you are in first, because the same topic is examined very differently in each.

In the US, the calculus-based probability course has a different number everywhere: MATH 461 Probability Theory or STAT 400 at Illinois, MATH 3235 Probability Theory or MATH 3215 Introduction to Probability and Statistics at Georgia Tech, MA 41600 Probability at Purdue, and STAT 134 Concepts of Probability at Berkeley. In the UK it usually starts in Year 1: Imperial lists Probability and Statistics, Manchester lists Probability I (MATH11711) and Statistics I (MATH11712), and Warwick lists ST120 Introduction to Probability.

In B.Tech programmes the subject is usually a second-year paper. AKTU's Mathematics-IV, for example, is titled 'PDE, Probability and Statistics' and covers correlation and regression, the binomial, Poisson and normal distributions, sampling, and t, Z and chi-square tests. The AICTE model curriculum includes a Probability and Statistics paper for CSE students. We work from your university's syllabus and past papers.

Topic map

Probability and statistics topics, with typical exam questions

This is the usual ground for a one- or two-semester course. The last column shows the kind of slip that costs marks most often in the scripts students share with us.

From probability rules to inference
TopicTypical questionWhere marks are lost
Probability rules and countingProbability of an event using combinations or the complementCounting ordered outcomes when order does not matter
Conditional probability and Bayes' theoremProbability of a cause given an observed resultConfusing P(A | B) with P(B | A)
Discrete random variablesFind the pmf, mean and variance of a countProbabilities that do not add to 1
Binomial, geometric and PoissonChoose the model and compute P(X ≤ k)Using the binomial when trials are not independent
Continuous random variablesFind k so that f(x) is a density; find P(a < X < b)Treating f(x) itself as a probability
Normal and exponential distributionsStandardise and read probabilities; the memoryless propertyDividing by the variance instead of the standard deviation
Expectation, variance and moment generating functionsE[g(X)], Var(aX + b), moments from an MGFWriting Var(aX + b) = a·Var(X) + b
Joint, marginal and conditional distributionsMarginals from a joint density; test for independenceWrong limits of integration over a non-rectangular region
Covariance and correlationCov(X, Y) and Var(X + Y)Assuming zero correlation means independence
Sampling distributions and the central limit theoremProbability that a sample mean exceeds a valueUsing σ instead of σ/√n
Point estimation and maximum likelihoodFind the MLE of a parameter; check biasDifferentiating the likelihood instead of its logarithm, then getting lost
Confidence intervalsA 95% interval for a mean or a proportionUsing z when σ is unknown and the sample is small
Hypothesis testsOne-sample and two-sample t-tests, chi-square tests, p-valuesStating the conclusion as 'H₀ is true'
Simple linear regressionLeast squares line, r and a predictionReading correlation as cause

Where students get stuck

The two jumps that make college probability feel harder than school

The first jump is from counting to calculus. At school, probability is mostly fractions, trees and Venn diagrams. At university, a continuous random variable is described by a density, probabilities are areas under it, and an expectation is an integral. Students who are comfortable with integration but have never seen it used this way often freeze at the first density question. Ajay links each new idea to a picture of the density, so that P(a < X < b) is visibly an area and E[X] is a balance point.

The second jump is from probability to inference. Students learn the normal distribution, then suddenly face sampling distributions, standard errors, intervals and tests, all in a few weeks. The usual problem is not arithmetic but meaning: what exactly is random in a confidence interval, what a p-value is the probability of, and why the sample size sits inside a square root. We slow down here and build the logic of one test fully, after which the others follow the same pattern.

Joint distributions are a smaller third hurdle. Most errors there are calculus errors: describing the region of integration, choosing the order of integration and setting the limits. Students who struggled with double integrals in multivariable calculus usually find the problem is the region, not the probability.

Worked examples

Three probability and statistics problems worked in full

Original problems at the level of a midterm or final, written out with each step a marker would look for.

A continuous random variable from its density

X has density f(x) = kx(2 − x) for 0 ≤ x ≤ 2, and 0 elsewhere. Find k, E[X], Var(X) and P(X > 1.5).

  1. A density must integrate to 1. ∫₀² x(2 − x) dx = [x² − x³/3] from 0 to 2 = 4 − 8/3 = 4/3, so k · 4/3 = 1 and k = 3/4.
  2. E[X] = (3/4) ∫₀² x²(2 − x) dx = (3/4)[2x³/3 − x⁴/4] from 0 to 2 = (3/4)(16/3 − 4) = (3/4)(4/3) = 1. This matches the symmetry of f about x = 1.
  3. E[X²] = (3/4) ∫₀² x³(2 − x) dx = (3/4)[x⁴/2 − x⁵/5] from 0 to 2 = (3/4)(8 − 32/5) = (3/4)(8/5) = 6/5.
  4. Var(X) = E[X²] − (E[X])² = 6/5 − 1 = 1/5.
  5. P(X > 1.5) = (3/4) ∫ from 1.5 to 2 of (2x − x²) dx = (3/4)[x² − x³/3] from 1.5 to 2 = (3/4)(4/3 − 9/8) = (3/4)(5/24) = 5/32.

Answer: k = 3/4, E[X] = 1, Var(X) = 1/5, P(X > 1.5) = 5/32 ≈ 0.156

Checking that E[X] agrees with the symmetry of the graph is a ten-second test that catches most integration slips.

The central limit theorem for a sample mean

Battery lifetimes have mean 40 hours and standard deviation 6 hours. A random sample of 36 batteries is tested. Find the approximate probability that the sample mean lifetime is below 38 hours.

  1. The lifetime distribution is not stated, but n = 36 is large enough for the central limit theorem: the sample mean X̄ is approximately normal.
  2. Mean of X̄ = 40. Standard error = σ/√n = 6/√36 = 6/6 = 1 hour.
  3. Standardise: z = (38 − 40)/1 = −2.
  4. P(X̄ < 38) ≈ P(Z < −2) ≈ 0.0228 from the standard normal table.

Answer: About 0.0228, roughly a 2.3% chance

Using σ = 6 instead of the standard error gives z = −0.33 and a probability near 0.37, a very different answer. The question is about the mean of 36 batteries, not one battery.

A confidence interval and a t-test on the same data

Ten measurements of a component's length (mm) are 52, 48, 55, 51, 49, 53, 54, 50, 56, 52. Assuming lengths are normal, find a 95% confidence interval for the mean and test H₀: μ = 50 against H₁: μ ≠ 50 at the 5% level.

  1. Sample mean: x̄ = 520/10 = 52.
  2. Deviations from 52: 0, −4, 3, −1, −3, 1, 2, −2, 4, 0. Their squares sum to 60, so s² = 60/9 ≈ 6.667 and s ≈ 2.582.
  3. Standard error: s/√10 ≈ 2.582/3.162 ≈ 0.8165. σ is unknown and n is small, so use the t distribution with 9 degrees of freedom; t₀.₀₂₅,₉ ≈ 2.262.
  4. 95% interval: 52 ± 2.262 × 0.8165 = 52 ± 1.847, which is about (50.15, 53.85).
  5. Test statistic: t = (52 − 50)/0.8165 ≈ 2.449. Since |t| > 2.262, reject H₀ at the 5% level; the two-sided p-value is about 0.037.
  6. Consistency check: 50 lies outside the 95% interval, which agrees with rejecting H₀: μ = 50 in a two-sided test at 5%.

Answer: 95% CI about (50.15, 53.85) mm; t ≈ 2.45, p ≈ 0.037, so reject H₀ at the 5% level.

Write the conclusion in context: there is evidence at the 5% level that the mean length differs from 50 mm. Never write that H₀ has been proved false.

Choosing the tool

Which distribution, which interval, which test

Exams rarely say which model to use. These are the questions Ajay teaches students to ask before writing anything.

  • A fixed number of independent trials, each a success or failure: binomial. Counts of events in an interval of time or space at a steady rate: Poisson. Waiting time for that kind of event: exponential.
  • A sum or mean of many independent values: think normal, through the central limit theorem, and work with the standard error.
  • One mean, σ known or a large sample: z interval or z-test. One mean, σ unknown and a small sample from a normal population: t with n − 1 degrees of freedom.
  • Two means: decide first whether the samples are paired or independent; a paired design is a one-sample test on the differences.
  • Counts in categories compared with expected counts, or a two-way table: chi-square, with the expected counts checked first.
  • Two quantitative variables: scatter plot, correlation and a least squares line, and a residual check before any prediction.

The matrix side of regression

Least squares, projections and the normal equations.

Linear algebra with Ajay

Regression connects to linear algebra: the least squares line comes from the normal equations, and multiple regression is the same idea with more columns. Students in data science courses often find statistics easier once they see it in matrix form.

How sessions work

What online probability and statistics sessions look like

Sessions are online, with handwritten working on a shared screen. Session times are agreed by your time zone when you get in touch.

1. Send your course first

Course title and code, syllabus or outline, the textbook or notes your course follows (for example Ross, Hogg and Tanis, or Walpole for engineers), a recent assignment or quiz you found hard, and your exam dates. If your course provides a formula sheet or statistical tables, send those too.

2. A diagnostic on two levels

One question tests calculus fluency (a density or an expectation) and one tests reasoning (interpreting an interval or a p-value). Together they show where to begin.

3. Each week

The idea from your latest lecture with a picture, one fully worked example, then questions you attempt while Ajay comments on both method and wording.

4. Before midterms and finals

Past papers under time, a model-choosing drill like the list above, and practice writing conclusions in context, which is where many marks are won or lost.

5. After results

We go through the marked paper and decide whether to stop, continue into the next course, or work on the statistics your project or major needs.

First month

Our usual first month for a student who is behind

This is a typical plan for a student who joins a few weeks into a semester. It is our usual approach, not a promise, and it moves with your syllabus.

Week 1: random variables

Discrete and continuous random variables, pmfs and densities, expectation and variance as sums and integrals.

Week 2: the standard distributions

Binomial, Poisson, exponential and normal: when each applies, how to compute with it, and the approximations between them.

Week 3: joint distributions and the CLT

Marginals, independence, covariance, then sums of random variables and the central limit theorem.

Week 4: into inference

Standard error, the first confidence interval and the first hypothesis test, with the logic made explicit.

Fit

Who these sessions are for, and who should look elsewhere

They suit college and university students in a calculus-based probability or statistics course, B.Tech students preparing the probability and statistics unit or clearing a back paper, and students in data science, economics or the sciences who want the theory behind the methods they use. Master's students who need the probability or statistics behind a course or project can book sessions on just that, as teaching only. Ajay holds an M.Tech and has 14+ years teaching maths.

They are not for anyone who wants an assignment or analysis done for them. GATE probability and statistics is taught by another Maths Bodhi tutor, not Ajay, so mention GATE when you enquire and we will put you in touch with that tutor. Home tuition is available in Gurgaon (Gurugram). Students elsewhere in India and abroad learn with us online.

Back paper in engineering maths

A plan for clearing a failed engineering mathematics paper.

Clearing a back paper

Academic integrity

Help to understand, never a substitute for your own work

We explain concepts, set practice and give feedback on attempts the student has made. We do not complete graded assignments, problem sets or data analysis projects, take online quizzes, tests or exams in anyone's place, or write reports, dissertations or theses that will be assessed. If you bring a graded question, Ajay teaches the method on a similar problem with different data, and you answer your own.

Study guides

Free guides for this course

Worked examples and exam advice from the Maths Bodhi team, topic by topic.

FAQ

Questions students ask about probability and statistics tutoring

Do I need strong calculus for university probability?

You need to integrate confidently, including over simple two-dimensional regions for joint distributions. If that is shaky, we fix it alongside the probability rather than separately.

Can you help with a mathematical statistics course with proofs?

Yes. We cover derivations such as expectations from MGFs, unbiasedness and maximum likelihood estimators, and the reasoning behind intervals and tests.

Which textbook do you follow?

The one your course uses. Send the chapter list and a problem set and Ajay keeps to your notation and tables.

My course uses R, Python or Excel. Is that covered?

We cover the statistics behind the output: what the numbers mean, which test was run and whether it was the right one. Sessions do not produce graded code or analyses.

Is B.Tech probability and statistics covered?

Yes. Distributions, joint distributions, correlation and regression, sampling and tests are regular topics in our engineering work. Send your university syllabus so we match the units.

What is the difference between a confidence interval and a hypothesis test?

They use the same standard error. A two-sided test at the 5% level rejects a value exactly when that value lies outside the 95% interval, as the third worked example shows.

I study in the UK. What time would sessions be?

IST is 5 hours 30 minutes ahead of the UK in winter and 4 hours 30 minutes ahead in summer, so 6 pm UK time is 11:30 pm or 10:30 pm IST. Times are agreed when you enquire.

How many sessions do students usually book?

Often one or two a week through a semester, and more in the weeks before a final. It depends on how far behind you are.

Is there a trial or demo class?

Yes, you can book a free demo class through the site. Bring a question from your own course.

Will you do my statistics assignment?

No. We teach the methods and review your own attempts; we never complete graded work.

Next step

Tell us about your probability or statistics course

Send your university, course name or paper code, textbook, a question you are stuck on and your time zone. Maths Bodhi will check whether Ajay is the right fit and suggest session times.