MBA(A) 205 Predictive Analytics and Big Data Notes cover all 4 units of the GGSIPU MBA (Analytics) Semester 3 paper in one 61-page printed book: 17 worked examples, 24 code listings (Python, HDFS, MapReduce, MongoDB and CQL) and 79 end-of-unit exam questions, mapped to the Scheme and Syllabus 2022-24 onwards.
Written by Prateek Talwar, Founder, Unnati Educations · Reviewed by Ritik Rawat, B.Tech, Senior Full Stack Developer · Checked against syllMBAAnalytics180225.pdf · Updated 14 October 2026 · Editorial policy
Scheme: MBA (Analytics) 2022-24 onwards | Applies to: batches on this scheme | Verified against: syllMBAAnalytics180225.pdf
MBA(A) 205 Predictive Analytics and Big Data at a glance: GGSIPU MBA (Analytics) Sem 3 credits, units and exam pattern
| Code and title | MBA(A) 205 Predictive Analytics and Big Data |
|---|---|
| Category and credits | Ability Enhancement · L 2, no T/P · 2 credits |
| Units and hours | 4 units, 28 hours (6, 8, 6, 8) |
| Internal assessment | Out of 25: Class Test-I 15 + presentation, viva-voce or group discussion 10 |
| End-term marks and pattern | Not printed in the official PDF |
| Applies to | Scheme and Syllabus 2022-24 onwards |
| Paired lab | None (MBA(A) 255 DBMS Lab pairs with MBA(A) 209) |
Half the hours go to modelling data, half to big data platforms; only Unit IV asks for "hands-on practical learning", though the paper has no practical credits. Examinations follow Ordinance 11.
It sits with the other GGSIPU MBA (Analytics) Semester 3 papers.
MBA(A) 205 Predictive Analytics and Big Data Notes unit-wise: 17 worked examples, 24 code listings and 79 exam questions across 4 units
Each unit of these GGSIPU MBA (Analytics) Sem 3 Predictive Analytics and Big Data notes is summarised below with page range and counts.
Every other paper's notes: GGSIPU MBA (Analytics) notes.
MBA(A) 205 Unit 1: Data Processing for Predictive Analysis
Pages 4 to 17. One income dataset of ten customers (₹4 lakh to ₹31 lakh) runs through min-max normalisation, z-score standardisation, decimal scaling, the ladder of powers up to Box-Cox, normality checks (Shapiro-Wilk p = 0.044 before the log, 0.996 after) and six outlier treatments. Counts: 5 worked examples, 4 Python listings, 6 figures, 17 exam questions.
| Basis | Z-score method | IQR method |
|---|---|---|
| Rule | |Z| > 3 | Outside Q1 − 1.5 IQR to Q3 + 1.5 IQR |
| Result on the data | Top z = 2.409, nothing flagged (masking) | Fences −3.25 and 24.75, ₹31 lakh flagged |
Quick look: z-score vs IQR on the notes' income data (from page 15)
- Basis
- Rule
- Z-score method
- |Z| > 3
- IQR method
- Outside Q1 − 1.5 IQR to Q3 + 1.5 IQR
- Basis
- Result on the data
- Z-score method
- Top z = 2.409, nothing flagged (masking)
- IQR method
- Fences −3.25 and 24.75, ₹31 lakh flagged
MBA(A) 205 Unit 2: Predictive Analytics
Pages 18 to 34. Every model runs on business data: a 20-store sales regression (R² 0.986), a 300-customer churn model, a 2-2-1 neural network with one back-propagation update, a 12-record insurance Naïve Bayes table and a sales Bayesian network, then evaluation with RMSE, MAPE, AUC 0.878 and a lift of 2.67. Counts: 10 worked examples, 4 Python listings, 7 figures, 23 exam questions.
| Variable | Odds ratio | Meaning |
|---|---|---|
| Charges | 1.008 per ₹1 | ₹100 more on the bill: odds × 2.14 |
| Tenure | 0.937 | Each extra month: odds about 6% lower |
| Complaints | 2.115 | Each extra complaint roughly doubles the odds |
Quick look: reading the churn model's odds ratios (from page 24)
- Variable
- Charges
- Odds ratio
- 1.008 per ₹1
- Meaning
- ₹100 more on the bill: odds × 2.14
- Variable
- Tenure
- Odds ratio
- 0.937
- Meaning
- Each extra month: odds about 6% lower
- Variable
- Complaints
- Odds ratio
- 2.115
- Meaning
- Each extra complaint roughly doubles the odds
MBA(A) 205 Unit 3: Introduction to Big Data & Analytics
Pages 35 to 45, the theory unit: Gartner and McKinsey definitions, three data types, seven Vs with Indian examples, a timeline from Codd (1970) to lakehouses, a 12-row BI versus big data table, three glossaries, the seven-step analytics flow and an 8-layer stack. Counts: 5 figures, 1 case study, 19 exam questions.
| Mode and tools | Results in |
|---|---|
| Batch (MapReduce, Spark, Hive) | Minutes to hours |
| Real-time (Kafka, Flink, Storm) | Milliseconds, e.g. card fraud alerts |
| Interactive (Spark SQL, Impala, Presto) | Quick ad-hoc answers |
Quick look: the three analysis modes (from page 42)
- Mode and tools
- Batch (MapReduce, Spark, Hive)
- Results in
- Minutes to hours
- Mode and tools
- Real-time (Kafka, Flink, Storm)
- Results in
- Milliseconds, e.g. card fraud alerts
- Mode and tools
- Interactive (Spark SQL, Impala, Presto)
- Results in
- Quick ad-hoc answers
MBA(A) 205 Unit 4: Big Data Analytics
Pages 46 to 60, also covered as "Big Data Analytics: Tools and Hands-on Practice". After NoSQL, CAP and BASE, and Hadoop with YARN, it runs a city-wise sales job as a Python mapper and reducer (Delhi 77,000, Mumbai 85,000, Pune 66,000), repeats it in PySpark, then covers MongoDB CRUD and a Cassandra keyspace in CQL. Counts: 16 code listings (8 with output), 2 worked examples, 20 exam questions.
| Phase | Output |
|---|---|
| Map (two input lines) | One (word, 1) pair per word: (big, 1), (data, 1), (is, 1) … |
| Shuffle and sort | big → [1, 1]; data → [1, 1]; is → [1, 1]; valuable → [1] |
| Reduce (sum) | (big, 2), (data, 2), (is, 2), (valuable, 1) |
Quick look: MapReduce word count, phase by phase (from page 52)
- Phase
- Map (two input lines)
- Output
- One (word, 1) pair per word: (big, 1), (data, 1), (is, 1) …
- Phase
- Shuffle and sort
- Output
- big → [1, 1]; data → [1, 1]; is → [1, 1]; valuable → [1]
- Phase
- Reduce (sum)
- Output
- (big, 2), (data, 2), (is, 2), (valuable, 1)
HDFS architecture and block replication in big data analytics: excerpt from page 49 of the MBA(A) 205 notes
Excerpt from page 49
HDFS, the storage layer of Hadoop, follows a master and slave design. Five parts work together to keep huge files safe on cheap machines.
NameNode (master)
It manages the file-system namespace and the metadata: file names, permissions, which blocks make up a file and on which DataNodes those blocks sit. It does not store the actual data.
DataNodes (slaves)
They store the real data blocks and serve read and write requests. Every 3 seconds each DataNode sends a heartbeat to the NameNode, along with block reports.
Secondary NameNode and Standby NameNode
The Secondary NameNode periodically merges the NameNode's edit log with the file-system image, a job called checkpointing. It is not a hot backup. In high-availability setups that role belongs to the Standby NameNode, which takes over automatically if the active NameNode fails.
Blocks, replication and rack awareness
A block is the smallest unit of storage: 128 MB by default (64 MB in Hadoop 1.x). Each block is kept as 3 replicas: one on the local node and two on nodes in a different rack, so the data survives even if a whole rack fails.
Worked illustration: storing a 500 MB sales log
With a 128 MB block size and a replication factor of 3, the file needs 500 ÷ 128 = 3.9, so 4 blocks: 128 + 128 + 128 + 116 MB. The last block uses only the space it needs. Total block replicas = 4 × 3 = 12, and raw storage used = 500 × 3 = 1,500 MB. If a DataNode fails, the NameNode notices the missing heartbeats and re-creates the lost replicas on other nodes.
How HDFS writes and reads
Write: the client asks the NameNode where to write, gets a list of DataNodes, writes the block to the first one, which pipelines it to the second and third; acknowledgements come back and the NameNode records the metadata. Read: the client asks the NameNode for block locations and reads each block directly from the nearest DataNode, switching to another replica if one fails.
How to write MBA(A) 205 answers: scaling numericals, model-output interpretation and hands-on command answers
MBA(A) 205 mixes three answer types, and the exam tips inside the MBA(A) 205 Predictive Analytics and Big Data IPU notes show how each should look. For scaling and outlier numericals, write the formula, then a working table. Say which SD you used: pandas divides by n − 1 and scikit-learn's StandardScaler by n, so ₹31 lakh becomes 2.409 or 2.539. Also note that the scaler is fitted on training data only; fitting it on the full data leaks test information into the model.
For Unit II model answers, interpret the number. If a model predicts ln(Sales) = 4.2, back-transform it (e^4.2 ≈ 66.7) before concluding. If you quote accuracy, add recall, precision and F1: the notes show a "no fraud" model that is 98% accurate on data with 2% fraud and still useless.
Worked answer: confusion matrix for the churn model (from page 31)
TP = 18, FN = 12, FP = 6, TN = 54 on 90 test customers. Accuracy = (18 + 54) ÷ 90 = 0.800; precision = 18 ÷ 24 = 0.750; recall = 18 ÷ 30 = 0.600; specificity = 54 ÷ 60 = 0.900; F1 = 2 × 0.75 × 0.60 ÷ 1.35 = 0.667. Close with the trade-off: at a 0.3 cut-off, recall rises to 0.833 but precision falls to 0.625, so more churners are caught with more false alarms.
For Unit III theory, the notes advise drawing the Vs diagram, giving each V an Indian example and ending with the summary table. For Unit IV hands-on questions, write the command, its purpose and the expected output; for MapReduce, show Map output, the shuffle and sort grouping and Reduce output on a small input such as word count, and add the SQL equivalent the notes give (GROUP BY for Map, SUM for Reduce).
Practise these formats on GGSIPU MBA (Analytics) important questions.
MBA(A) 205 exam corner: 56 short (two to five marks) and 23 long (ten to fifteen marks) questions by unit
| Unit | Short (2–5 marks) | Long (10–15 marks) | Worked examples / code listings |
|---|---|---|---|
| I | 12 | 5 | 5 / 4 |
| II | 17 | 6 | 10 / 4 |
| III | 13 | 6 | 0 / 0 |
| IV | 14 | 6 | 2 / 16 |
MBA(A) 205 exam questions and worked material per unit
- Unit
- I
- Short (2–5 marks)
- 12
- Long (10–15 marks)
- 5
- Worked examples / code listings
- 5 / 4
- Unit
- II
- Short (2–5 marks)
- 17
- Long (10–15 marks)
- 6
- Worked examples / code listings
- 10 / 4
- Unit
- III
- Short (2–5 marks)
- 13
- Long (10–15 marks)
- 6
- Worked examples / code listings
- 0 / 0
- Unit
- IV
- Short (2–5 marks)
- 14
- Long (10–15 marks)
- 6
- Worked examples / code listings
- 2 / 16
One exam-style question per unit, with pointers:
- Unit I: How is Pearson's coefficient of skewness calculated? 3 × (Mean − Median) ÷ SD; on the income data 3 × (12 − 9.5) ÷ 7.888 = 0.951, right-skewed (page 10).
- Unit II: Differentiate between R² and adjusted R². Adding the irrelevant StaffAge keeps R² at 0.986 while adjusted R² slips from 0.983 to 0.982 (pages 21-22).
- Unit III: Differentiate between schema-on-write and schema-on-read. A warehouse fixes structure before loading; big data applies it at analysis time (page 39).
- Unit IV: Calculate QUORUM for RF = 5. ⌊5 ÷ 2⌋ + 1 = 3 replicas; the notes work RF = 3 on page 57.
The MBA(A) 205 notes list 56 short-answer and 23 long-answer exam questions across 4 units, plus 17 worked examples and 24 code listings.
More practice sits in the MBA(A) 205 important questions.
MBA(A) 205 and the GGSIPU MBA (Analytics) Sem 3 papers it connects to
MBA(A) 205 has no paired lab; its tool practice sits inside Unit IV. All Sem 3 papers are in the GGSIPU MBA (Analytics) syllabus.
Unit II's regression and back-propagation meet MBA(A) 207 Artificial Intelligence and Machine Learning. Unit IV's RDBMS versus NoSQL table, ACID versus BASE and the SQL equivalents beside each MapReduce and MongoDB job lean on MBA(A) 209 Database Management System and MBA(A) 255 DBMS Lab.
The rest of Sem 3: MBA(A) 201 Strategic Management, MBA(A) 203 Entrepreneurial Development & Start-Ups, MBA(A) 211 Design Thinking and Innovation, MBA(A) 213 Summer Training Project Report, MBA(A) 233 MOOCs II/Open Elective, and two electives from the HR, Marketing or Finance stream.
Who writes and checks the MBA(A) 205 Predictive Analytics notes?
- Prateek Talwar, author: Founder, Unnati Educations · ACS, ICSI (membership no. 58704, since 15 May 2019) · B.Com, University of Delhi (SOL), 2012 · bio
- Ritik Rawat, subject reviewer: Senior Full Stack Developer, Unnati Educations · B.Tech · checks every code listing against its printed output · bio
These Predictive Analytics and Big Data notes in English follow the 2022-24 onwards scheme unit by unit. The 2026-27 edition uses one running dataset per method family and cites India's DPDP Act, 2023 under data governance. Last updated 14 October 2026; see our editorial policy.
The same team prepares GGSIPU important questions for every course it covers.
Do the MBA(A) 205 notes cover the full GGSIPU syllabus?
Yes. Every topic under the four official units of MBA(A) 205 has its own section in the notes; the unit check found no gap.
| Official unit | Official topics | Where in the notes | Status |
|---|---|---|---|
| I: Data Processing for Predictive Analysis | Data Transformation, Min-Max Normalization, Z-Score Standardization, Transformations to Achieve Normality, Graphical and Numerical Methods for identifying outliers | Sections 2-8, pages 5-16 | Covered |
| II: Predictive Analytics | Multiple Regression and Model Building, Logistic Regression, Neural Networks, Naïve Bayes and Bayesian Networks, Model Evaluation Techniques | Sections 2-7, pages 18-33 | Covered |
| III: Introduction to Big Data & Analytics | What is Big Data? Characteristics and Evolution, Traditional BI Versus Big Data, Terminologies, Analytics Flow, Big Data Stack | Sections 1-7, pages 35-44 | Covered |
| IV: Big Data Analytics | NoSQL, Hadoop, MapReduce, MongoDB and Cassandra, hands-on practical learning on these tools | Sections 2-7, pages 46-59 | Covered (HDFS commands, Hadoop Streaming, mongosh, CQL) |
MBA(A) 205 official syllabus mapped to the notes
- Official unit
- I: Data Processing for Predictive Analysis
- Official topics
- Data Transformation, Min-Max Normalization, Z-Score Standardization, Transformations to Achieve Normality, Graphical and Numerical Methods for identifying outliers
- Where in the notes
- Sections 2-8, pages 5-16
- Status
- Covered
- Official unit
- II: Predictive Analytics
- Official topics
- Multiple Regression and Model Building, Logistic Regression, Neural Networks, Naïve Bayes and Bayesian Networks, Model Evaluation Techniques
- Where in the notes
- Sections 2-7, pages 18-33
- Status
- Covered
- Official unit
- III: Introduction to Big Data & Analytics
- Official topics
- What is Big Data? Characteristics and Evolution, Traditional BI Versus Big Data, Terminologies, Analytics Flow, Big Data Stack
- Where in the notes
- Sections 1-7, pages 35-44
- Status
- Covered
- Official unit
- IV: Big Data Analytics
- Official topics
- NoSQL, Hadoop, MapReduce, MongoDB and Cassandra, hands-on practical learning on these tools
- Where in the notes
- Sections 2-7, pages 46-59
- Status
- Covered (HDFS commands, Hadoop Streaming, mongosh, CQL)
Check the official MBA (Analytics) Scheme and Syllabus 2022-24 onwards (PDF), printed page 55. The notes add decimal scaling, robust scaling, Spark and the Lambda and Kappa architectures.
Every course in GGSIPU notes is mapped to its official PDF the same way.
MBA(A) 205 Predictive Analytics notes FAQs
How do I get the full MBA(A) 205 notes?
Message Unnati Educations on WhatsApp at 9899436384 using the "Get Full Notes" button. The MBA(A) 205 notes come only as a 61-page printed book; they are not available as a PDF. Delivery takes a maximum of 5 to 7 days, depending on your location. The button pre-fills your message with the paper code and a reference number.
How are MBA(A) 205 marks split, and what is the passing mark?
The official MBA (Analytics) PDF prints internal assessment for MBA(A) 205 out of 25: Class Test-I for 15 and a presentation, viva-voce or group discussion for 10. It does not print the end-term marks, the total or a pass percentage. The scheme follows GGSIPU Ordinance 11, which sets end-term minimums and re-appear rules, so check those on ipu.ac.in.
Which MBA(A) 205 unit has the most hands-on code?
Unit IV. It carries 16 of the 24 code listings in the MBA(A) 205 notes: HDFS shell commands, a Python mapper and reducer run through Hadoop Streaming, the same job in PySpark, MongoDB insert, find, update, delete and aggregation, and CQL to create and query Cassandra tables. Eight of those show printed output, and Unit III has none, being pure theory.
Do I need Python or Hadoop installed to study the MBA(A) 205 notes?
No. In the MBA(A) 205 notes, 15 of the 24 listings have their output printed beside them, so most of the logic can be followed on paper. To practise, the notes suggest a single-node Hadoop on a laptop, a virtual machine or Docker, or a cloud service such as AWS EMR, Azure HDInsight or Google Dataproc.
Do the MBA(A) 205 notes match the 2022-24 onwards scheme?
Yes. The MBA(A) 205 notes follow the MBA (Analytics) Scheme and Syllabus 2022-24 onwards, whose Semester 3 papers were approved by the Academic Council on 19.07.2023. All four official units and every topic under them have a matching section, from min-max normalisation in Unit I to hands-on Cassandra in Unit IV. The syllabus map above gives the pages.
Are the MBA(A) 205 exam questions answered in the notes?
The 79 MBA(A) 205 exam questions are listed at the end of each unit without separate model answers; each is answered by that unit's sections. For example, the Unit I long numerical on the data 4, 6, 7 … 31 is worked step by step inside the unit, from min-max values and z-scores to the IQR fences.
Get the full MBA(A) 205 Predictive Analytics and Big Data Notes on WhatsApp
| Unit | Pages | Page count |
|---|---|---|
| I | 4-17 | 14 |
| II | 18-34 | 17 |
| III | 35-45 | 11 |
| IV | 46-60 | 15 |
Pages per unit in the MBA(A) 205 printed book
- Unit
- I
- Pages
- 4-17
- Page count
- 14
- Unit
- II
- Pages
- 18-34
- Page count
- 17
- Unit
- III
- Pages
- 35-45
- Page count
- 11
- Unit
- IV
- Pages
- 46-60
- Page count
- 15
61 pages in all, with 21 figures, 3 case studies and a quick-revision table per unit. Printed book only, not available as a PDF; delivery takes a maximum of 5 to 7 days, depending on your location.
Order the printed MBA(A) 205 book on WhatsApp (4 units, 61 pages)
Unnati Educations is an independent education provider based in Vikaspuri, Delhi. We are not affiliated with, authorised by, or endorsed by Guru Gobind Singh Indraprastha University (GGSIPU). 'GGSIPU' and 'IPU' are used only to describe the syllabus our notes follow. For official information, visit ipu.ac.in.
Ready for the full notes?
Discuss coverage and delivery with our team.