Quick answer. Nine units need roughly eight books plus two self assembled areas. Gupta and Kapoor for the classical core, Rohatgi and Saleh for inference depth, Medhi for stochastic processes, Anderson or Johnson and Wichern for multivariate, Gujarati for econometrics, Brockwell and Davis for time series, and an Introduction to Statistical Learning for the machine learning unit. Units 8 and 9 have no standard examination book, and that is exactly why most candidates leave them until it is too late.
The Reality of Booklists for This Exam
Almost every RBI DSIM booklist online is a recycled UPSC ISS or UGC NET list. Those lists cover Units 1 to 6 reasonably and then stop.
Units 7, 8 and 9 are roughly a third of the syllabus, they are where most Statistics postgraduates have genuine gaps, and no recycled list covers them.
This page covers all nine.
Unit Wise Mapping
| Unit | Primary resource | Status |
|---|---|---|
| 1. Probability, Distributions, Sampling | Gupta and Kapoor | Essential |
| 2. Linear Models and Economic Statistics | Montgomery Peck Vining, plus Gupta and Kapoor for index numbers | Essential |
| 3. Statistical Inference | Rohatgi and Saleh | Essential |
| 4. Stochastic Processes | J. Medhi | Essential |
| 5. Multivariate Analysis | Johnson and Wichern | Essential |
| 6. Econometrics and Time Series | Gujarati, plus Brockwell and Davis | Essential |
| 7. Optimization and Statistical Computing | Assemble it. See below. | Gap |
| 8. Data Science and Machine Learning | An Introduction to Statistical Learning | Essential |
| 9. Database and Data Warehouse | Assemble it. See below. | Gap |
| Paper III English | Any competitive English writing book | Light |
Unit 1: Probability, Distributions and Sampling Theory
Fundamentals of Mathematical Statistics, S.C. Gupta and V.K. Kapoor
Your core theory base for probability, standard distributions and exact sampling distributions.
Read: theory of probability, random variables and distribution functions, mathematical expectation and generating functions, theoretical discrete and continuous distributions, sampling distributions of chi square, t and F.
Solve the exercises. Reading Gupta and Kapoor without working through the problems is the most common wasted month in Indian Statistics preparation.
For the sampling half
Gupta and Kapoor covers sampling methods lightly. Cochran’s Sampling Techniques or Singh and Chaudhary gives you stratified, systematic, cluster, two stage and PPS sampling with ratio and regression estimation properly.
Note the syllabus names characteristic functions, and the Logistic, Log normal and Weibull distributions. Check your edition covers all three.
Unit 2: Linear Models and Economic Statistics
Introduction to Linear Regression Analysis, Montgomery, Peck and Vining
Read: simple and multiple linear regression, Gauss Markov setup, diagnostic checks, polynomial regression, Box Cox and log transformations, weighted least squares, multicollinearity, model selection criteria, residual diagnostics, dummy variables and outlier detection.
Also covers the shrinkage models the syllabus names: ridge regression, LASSO and Elastic Net.
For the linear algebra portion
The syllabus names g inverse, inverse of partitioned matrices, idempotent matrices and simultaneous reduction of two quadratic forms. Those are linear models topics rather than general linear algebra.
A linear models text such as A.M. Kshirsagar, or Rencher and Schaalje, covers them. A general Schaum’s Linear Algebra will not.
For index numbers and inequality
Gupta and Kapoor covers index number construction, chain base conversion, base shifting, splicing and deflating.
Gini coefficient and Lorenz curves need an economic statistics or development economics source. Any standard text on measurement of inequality works, and the concepts are short.
Unit 3: Statistical Inference
An Introduction to Probability and Statistics, Rohatgi and Saleh
Read: unbiasedness, sufficiency, consistency, efficiency, factorization theorem, complete statistics, MVUE, Rao Blackwell, Lehmann Scheffe, Cramer Rao, methods of estimation, Neyman Pearson theory, likelihood ratio tests, goodness of fit.
Why this over Casella and Berger. Rohatgi and Saleh sits at the right level for an Indian postgraduate examination and covers Bartlett’s test and the non parametric range the syllabus names.
For the non parametric portion
Gibbons and Chakraborti covers Kolmogorov Smirnov, sign test, Wilcoxon signed rank and rank sum, Mann Whitney, Kruskal Wallis, Friedman’s test, Kendall’s tau and Spearman.
Friedman’s test appears in the DSIM syllabus and is absent from several other Statistics examinations, so check you have it.
Kernel density estimation appears at the end of Unit 3. Most inference books do not cover it. It is also in Unit 8, so cover it there.
Unit 4: Stochastic Processes
Stochastic Processes, J. Medhi
Read: Poisson process with arrival, interarrival and conditional arrival distributions, non homogeneous processes, law of rare events, compound Poisson processes, Markov chains with transition matrices, Chapman Kolmogorov equations, regular chains, stationary distributions, periodicity and limit theorems, Brownian motion and martingales.
Alternative: Sheldon Ross, Introduction to Probability Models, which is more accessible for Markov chains and the Poisson process.
Martingales are named in the syllabus and are thinly covered in Ross. Medhi handles them better.
Unit 5: Multivariate Analysis
Applied Multivariate Statistical Analysis, Johnson and Wichern
Read: multivariate normal distribution and properties, Mahalanobis D squared, linear discriminant analysis, canonical correlation, principal components analysis, factor analysis and cluster analysis.
Note what is different here. For other Statistics examinations we tell students to skip PCA, discriminant analysis and clustering. For DSIM they are explicitly in the syllabus. Read them.
Logit and probit models also sit in this unit. Johnson and Wichern is thin there, so take them from Gujarati, where they are covered properly.
Unit 6: Econometrics and Time Series
This unit needs two books.
Basic Econometrics, Damodar Gujarati
Read: general linear model, OLS and GLS, heteroscedasticity, autocorrelation and its tests, multicollinearity, ridge regression, instrumental variables, panel regression, distributed lag models, simultaneous equations, identification, rank and order conditions, and estimation methods for simultaneous equations.
Also read logit and probit here for Unit 5.
Free alternative: Professor Shalabh’s econometrics lecture notes from IIT Kanpur are exam friendly, thorough and cost nothing. Many candidates find them tighter than Gujarati for revision.
Introduction to Time Series and Forecasting, Brockwell and Davis
Read: stationarity, AR, MA and ARMA processes, model identification from ACF and PACF, estimation and diagnostics, Box Jenkins methodology, ARIMA and SARIMA, order of integration, trend stationary versus difference stationary, tests of non stationarity.
ARCH and GARCH are named in the syllabus and are thinly covered in Brockwell and Davis. Take them from a financial econometrics source or from Gujarati’s chapter on volatility.
Unit 7: Optimization and Statistical Computing
No single book covers this unit. Assemble it.
| Half | Resource |
|---|---|
| Optimization | Any operations research text for linear programming and the simplex method. For Newton, gradient, conjugate gradient and quasi Newton methods, a numerical optimization text or good lecture notes. |
| Statistical computing | Statistical Computing with R by Maria Rizzo, or Computational Statistics by Givens and Hoeting |
Givens and Hoeting is the closest single match for the second half. It covers EM algorithm, bootstrap, jack knife, cross validation, MCMC, Gibbs sampling, Metropolis Hastings and simulated annealing, which is almost exactly the syllabus text.
This is the unit candidates most underestimate. MCMC, Gibbs sampling and Metropolis Hastings are research level topics taught briefly, if at all, in most Master’s programmes. Give this unit real time and start it early.
Unit 8: Data Science, AI and Machine Learning
An Introduction to Statistical Learning, James, Witten, Hastie and Tibshirani
This is the right book for this unit, and it is free to download legally from the authors.
Read: linear and logistic regression, penalized regression with ridge and LASSO, naive Bayes, k nearest neighbours, decision trees, random forests, boosting, support vector machines, unsupervised learning with principal components and clustering, cross validation and model selection.
Why this book. It is written for people with a statistics background rather than a computer science one, and its notation will feel familiar to you.
For what it does not cover
Recurrent neural networks, convolutional neural networks and natural language processing go beyond it.
Use Analytics Vidhya or GeeksforGeeks selectively, one topic at a time. Do not attempt a full deep learning course. The syllabus asks for the concepts, not implementation.
The practical machine learning list in the syllabus, recursive feature elimination, variance inflation factor, ensembling and stacking, grid search, feature importance and cross validation strategies, is best learned as short focused reading rather than from a textbook.
Unit 9: Database and Data Warehouse Management
No examination book exists for this unit either. Assemble it.
| Topic | Source |
|---|---|
| RDBMS fundamentals, normalization | Any standard DBMS textbook, or a good online course |
| SQL and joins | Practise on a free SQL platform rather than reading about it |
| NoSQL types | Short conceptual reading is enough |
| Star and snowflake schemas, ETL, OLAP versus OLTP | Data warehousing notes or a short online course |
The most efficient approach. Spend a weekend actually writing SQL queries on a free practice site. Inner, left, right and outer joins are learned in an hour of doing and never learned by reading.
Everything else in this unit is definitional. Build one page of notes covering normalization forms, the four join types, NoSQL categories, star versus snowflake, ETL and OLAP versus OLTP. That single page is most of the unit.
Paper III: English
Any standard competitive English writing book. Practise precis writing to a strict word limit, essays on economic and data themes, and reading comprehension.
Build typing speed. This paper is typed on a keyboard, and slow typing costs marks that have nothing to do with your English.
The Minimum Viable List
If you buy nothing else:
- Fundamentals of Mathematical Statistics, Gupta and Kapoor
- An Introduction to Probability and Statistics, Rohatgi and Saleh
- Basic Econometrics, Gujarati
- Stochastic Processes, J. Medhi
- Applied Multivariate Statistical Analysis, Johnson and Wichern
- An Introduction to Statistical Learning, free download
- Computational Statistics, Givens and Hoeting
- Introduction to Time Series and Forecasting, Brockwell and Davis
Plus self assembled notes for Unit 9 and the deep learning portion of Unit 8.
Depth beats collection. Every extra book you buy is time stolen from revising the one you already own.
Order of Study
| Stage | Units | Why |
|---|---|---|
| 1 | 1, 2, 3 | The classical core. Fastest for a Statistics postgraduate. |
| 2 | 8, 9 | Do these second, not last. They are new for almost everyone and they do not compress. |
| 3 | 4, 5 | Stochastic processes and multivariate |
| 4 | 6 | Econometrics and time series, the largest unit |
| 5 | 7 | Optimization and computing |
| Throughout | Paper III English, and descriptive writing practice by hand |
Units 8 and 9 sit at stage 2 deliberately. The instinct is to leave the unfamiliar material for later. That instinct is why most candidates arrive at the examination able to answer six units out of nine.
Frequently Asked Questions
How many books do I need for RBI DSIM?
Around eight, plus self assembled material for Unit 9 and the deep learning portion of Unit 8.
Which book covers machine learning for RBI DSIM?
An Introduction to Statistical Learning by James, Witten, Hastie and Tibshirani. It is free to download legally from the authors and is written for readers with a statistics background.
Is there a book for the RBI DSIM database unit?
No standard examination book exists. Use any DBMS text for fundamentals, practise SQL joins on a free platform, and build one page of notes on normalization, NoSQL types, star and snowflake schemas, ETL and OLAP versus OLTP.
Which book covers MCMC and Gibbs sampling?
Computational Statistics by Givens and Hoeting is the closest single match for Unit 7, covering the EM algorithm, bootstrap, cross validation, MCMC, Gibbs sampling, Metropolis Hastings and simulated annealing.
Is Gujarati enough for the econometrics unit?
For the econometrics half, largely yes. The time series half needs a dedicated book such as Brockwell and Davis, and ARCH and GARCH need a separate source.
Can I use my UPSC ISS books for RBI DSIM?
For Units 1 to 6, substantially yes. Units 7, 8 and 9 have no ISS equivalent and need their own resources.
Are free resources enough for RBI DSIM?
For some units, genuinely yes. An Introduction to Statistical Learning is free, Professor Shalabh’s econometrics notes from IIT Kanpur are free, and SQL is best learned on free practice platforms. The classical Statistics units still need proper textbooks.