LAISHA
Open Laisha

ICAR-IASRI MSc and PhD Statistical & Machine Learning for Agriculture: practice questions

ICAR-IASRI MSc and PhD (Agricultural Statistics, Computer Application, Bioinformatics) · India · ICAR-Indian Agricultural Statistics Research Institute with the IARI Graduate School

Original practice questions written to the published syllabus · not past papers
10worked here, free
94written to this exam in this subject
11syllabus units

About ICAR-IASRI MSc and PhD

ICAR-IASRI MSc and PhD (Agricultural Statistics, Computer Application, Bioinformatics) is a postgraduate admission exam run by ICAR-Indian Agricultural Statistics Research Institute with the IARI Graduate School, for candidates in India. Admission through ICAR AIEEA-PG (Master's) and AICE-JRF/SRF (PhD) plus the Graduate School's own process.

Official site: iasri.icar.gov.in.

We have not been able to confirm these details against the exam body's own specification, so fees, dates and pass marks are left out on purpose. Check the current format, eligibility, fees and dates on the official site.

Statistical & Machine Learning for Agriculture in ICAR-IASRI MSc and PhD

Statistical learning with agricultural data: penalised regression and genomic prediction, ensembles and SVMs for yield and remote sensing, CNNs for plant images, and blocked validation.

Prediction from data with honest validation: the bias-variance trade-off, regularised regression, trees and ensembles, support vector machines, neural networks and deep learning for images, unsupervised learning, and how to use them on yield, disease, price and image data without fooling yourself. The Agribusiness and Agricultural Engineering tracks introduce ML; this is the statistician's full course.

Our 94 ICAR-IASRI MSc and PhD questions in this subject: 26 easy, 52 medium, 16 hard, across 11 units.

10 practice questions, worked

Choose an answer, then open the working. Each explanation says why the right answer is right and why the others are not.

These 10 were written to ICAR-IASRI MSc and PhD and are not shown on any other exam's page.

Question 1 · easy

A model predicts whether a leaf image shows rust or not. What kind of learning task is this?

  1. Dimensionality reduction of the image pixels
  2. Supervised regression on a numerical outcome
  3. Supervised classification
  4. Unsupervised clustering of leaf images into groups
Answer and working

C. Supervised classification

The outcome (rust or not) is a known category for training images, so this is supervised classification. Regression needs a numerical outcome; clustering has no outcome.

Question 2 · hard

Ridge regression of a trait on thousands of standardised marker genotypes is essentially equivalent to which genomic prediction model?

  1. Principal component regression
  2. Marker-assisted backcrossing
  3. GBLUP
  4. Single-marker t-tests
Answer and working

C. GBLUP

Ridge regression on markers and GBLUP (BLUP with a genomic relationship matrix) are equivalent models with an appropriate penalty.

Question 3 · medium

A screening model for a rare cattle disease gives 45 true positives, 5 false negatives, 30 false positives and 920 true negatives. What is its precision?

  1. 0.90
  2. 0.97
  3. 0.60
  4. 0.965
Answer and working

C. 0.60

Precision = TP/(TP + FP) = 45/75 = 0.60. 0.90 is sensitivity; 0.97 specificity; 0.965 accuracy, which looks high because negatives dominate.

Question 4 · easy

What does an area under the ROC curve (AUC) of 0.5 indicate?

  1. Half of positives correctly classified
  2. Perfect separation of the classes
  3. A threshold of 0.5 is optimal
  4. Ranking no better than chance
Answer and working

D. Ranking no better than chance

AUC is the probability that a random positive scores higher than a random negative; 0.5 is chance and 1 is perfect ranking.

Question 5 · medium

In epsilon-insensitive support vector regression, which training points become support vectors?

  1. Only the point with the largest residual
  2. Those inside the epsilon tube
  3. Those on or outside the epsilon tube
  4. Every training point
Answer and working

C. Those on or outside the epsilon tube

Errors smaller than epsilon are ignored; points on or outside the tube carry non-zero weights and define the fit.

Question 6 · medium

How many parameters (weights plus biases) does a fully connected layer from 100 inputs to 50 units have?

  1. 5,100
  2. 5,000
  3. 5,050
  4. 150
Answer and working

C. 5,050

100 x 50 = 5,000 weights plus 50 biases = 5,050. 5,000 omits biases; 5,100 counts 100 biases.

Question 7 · medium

Validation error rises after epoch 20 while training error keeps falling. Which technique uses this signal?

  1. Batch normalisation
  2. Adding more convolutional layers
  3. Increasing the learning rate
  4. Early stopping
Answer and working

D. Early stopping

Early stopping halts training (or keeps the weights) where validation error is lowest, limiting overfitting.

Question 8 · medium

Why is k-means usually run from many random starting centroids?

  1. Each start uses a different k
  2. Random starts remove the need to standardise
  3. It cannot converge from a single start
  4. It converges only to a local minimum
Answer and working

D. It converges only to a local minimum

Lloyd's algorithm decreases the within-cluster sum of squares until it stops, but the solution depends on the start; the best of many runs is kept.

Question 9 · medium

Why can principal component regression miss an important predictor signal?

  1. Components are always correlated with each other
  2. Principal components cannot be used in regression
  3. Components are chosen without looking at y
  4. It uses too many components by definition
Answer and working

C. Components are chosen without looking at y

A low-variance component may carry the information about y; partial least squares builds components using covariance with y.

Question 10 · medium

When the rainfall predictor is randomly shuffled in held-out data, a model's RMSE rises from 0.40 to 0.55 t/ha. What is rainfall's permutation importance?

  1. 0.15 t/ha
  2. 0.55 t/ha
  3. 1.38 t/ha
  4. 0.95 t/ha
Answer and working

A. 0.15 t/ha

Permutation importance is the increase in error: 0.55 - 0.40 = 0.15. 1.38 is the ratio of the errors, an alternative scale.

A study plan for Statistical & Machine Learning for Agriculture in ICAR-IASRI MSc and PhD

This plan follows where our questions actually fall, so the time goes where the marks are likely to be. It is our reading of the syllabus, not the exam body's weighting.

  1. First, the heavy units (39% of the 94 questions): Neural networks and deep learning; CNNs for plant and field images (10, 11%); Cross-validation: k-fold and leave-one-out; spatial and temporal blocking (9, 10%); Decision trees, random forests and gradient boosting (9, 10%); Linear and quadratic discriminant analysis and logistic classification (9, 10%).

    Read the manual chapters for these first and do every worked question you can find on them.

  2. Then the middle band: Regularised regression: ridge, lasso and elastic net (9, 10%); Dimensionality reduction: PCA and embeddings (8, 9%); k-nearest neighbours and naive Bayes (8, 9%); Model interpretation: variable importance and SHAP (8, 9%).

    Aim for one sitting per unit, then mixed questions across them.

  3. Then the lighter units: Supervised learning and the bias-variance trade-off (8, 9%); Support vector machines (8, 9%); Unsupervised learning: k-means, k-medoids and hierarchical clustering (8, 9%).

    Learn the definitions, classifications and standard numbers; these units are more often recalled than worked.

  4. Leave time for the hard ones: Regularised regression: ridge, lasso and elastic net (3 hard of 9), k-nearest neighbours and naive Bayes (2 hard of 8), Model interpretation: variable importance and SHAP (2 hard of 8).

    Come back to these after a first pass, untimed, and read every explanation, including for the options you rejected.

  5. Bank the easy marks: k-nearest neighbours and naive Bayes (3 easy of 8); Supervised learning and the bias-variance trade-off (3 easy of 8).

    Do these early in revision so they are fast on the day.

  6. Finish timed. Sit the subject as a timed set without notes, mark it, and send each wrong answer back to its unit above.

Last updated · How Laisha writes and checks its pages · Report a mistake

Ask Laisha about this

Opens Laisha's chat with your question ready to send. Sign in or start free.

Unlock every question — ICAR-IASRI MSc and PhD Statistical & Machine Learning for Agriculture

US$10 a month · one price worldwide · cancel any time

  • Every practice question in this subject, with full working
  • Timed exam papers and progress by unit
  • Ask Laisha to explain any answer — up to 1,000 questions a month
Free or Smarter
FreeSmarter
Grow guides, pest pages, articlesRead allRead all
Ask Laisha5 a dayUp to 1,000 a month
Exam practiceThe sample on each pageEvery question, timed papers
Crop calendar and pest alerts—For your own fields
Calculators and tools—All
Uploads and reports—100 a month, real files
10 days free for every new account, no card.