DATA SCIENCE (AD-404) Unit 2 Notes - RGPV

How unit 2 is examined

This unit covers unstructured data, the four types of analytics, data visualization (box plot, histogram, scatter plot, feature map, t-SNE) and Advanced Excel; the marks sit in importance of unstructured data, the analytics types, data validation and macros.

Importance of Unstructured Data

Medium weight

Definition. Unstructured data is data with no predefined data model or row-column schema, such as text, images, audio, video and social media posts. It makes up roughly 80 percent of all data generated and holds insight that tables alone cannot give.

Key points.

Its sources are emails, documents, web pages, social media posts, images, audio calls, video and sensor logs.
Its volume is huge and grows faster than structured data, so ignoring it means ignoring most of what an organisation knows.
It gives business value, for example customer opinion from reviews, complaint themes from call transcripts and brand sentiment from tweets.
In healthcare, doctors' notes, X-rays and MRI scans support diagnosis and patient-risk prediction.
In finance, news, filings and call transcripts support fraud detection, credit scoring and market sentiment analysis.
In social media, posts and comments support trend detection, recommendation and targeted advertising.
Its challenges are storage of large volumes, no fixed schema, difficult search, noise and privacy concerns.
It is processed using NLP, computer vision, text mining and deep learning, on tools such as Hadoop, Spark and NoSQL stores such as MongoDB.

Answer frame. Open with the definition and the 80 percent figure; then sources (1-2), value with healthcare, finance and social media examples (3-6); then challenges and tools (7-8); close with one line that unstructured analytics turns raw text and media into decisions.

Asked: [7 marks] (Dec 2024, Jun 2025, Jun 2026) Write a brief note on importance of unstructured data / its use in healthcare, finance and social media / importance of Unstructured Data Analytics in modern organizations with examples.

Unstructured Data Analytics: Descriptive, Diagnostic, Predictive and Prescriptive

Definition. Analytics is the process of examining data to draw conclusions, and it has four levels that answer four questions: what happened, why, what will happen and what should we do. Descriptive looks back, diagnostic explains, predictive forecasts and prescriptive recommends.

Type	Question	Techniques	Case example (retail reviews)
Descriptive	What happened?	Mean, median, mode, spread, charts, word counts	30 percent of this month's reviews are negative
Diagnostic	Why did it happen?	Drill-down, correlation, root-cause analysis	Negative reviews cluster around late delivery
Predictive	What will happen?	Regression, classification, time series, ML	Customers with such reviews will likely churn
Prescriptive	What should we do?	Optimisation, simulation, recommendation rules	Switch courier and offer a discount to at-risk customers

Key points.

Descriptive analysis summarises past data using central tendency (mean, median, mode), dispersion (range, variance, standard deviation) and visualization, and it is the starting point of all analysis.
Diagnostic analysis digs into causes by comparing groups, drilling down and finding correlations.
Predictive analysis uses historical patterns with statistical or machine learning models to forecast future outcomes.
Prescriptive analysis goes further and recommends the best action using optimisation and what-if simulation.
Value and difficulty both rise from descriptive to prescriptive.

Answer frame. Descriptive (Jun 2023): define, give measures of centre and spread, name charts, add a small example. Predictive and prescriptive (Jun 2024): define each, give techniques, then a use-case contrast such as forecast demand versus decide stock to order.

Asked: [7 marks] (Jun 2023) Briefly explain about descriptive data analysis. Asked: [6 marks] (Jun 2024) Explain the following data analysis methods: i) Predictive ii) Prescriptive.

Data Visualization: Boxplots

Definition. Data visualization is the graphical presentation of data so that patterns, trends and outliers are seen quickly and decisions are easier. A box plot displays the five-number summary: minimum, $Q_1$, median, $Q_3$ and maximum.

 o          |----[   Q1 | median |  Q3   ]-----|
outlier  whisker           box (IQR)        whisker

Key points.

The box spans $Q_1$ to $Q_3$, so its length is $IQR = Q_3 - Q_1$, and the line inside it is the median.
Whiskers extend to the smallest and largest values within $Q_1 - 1.5\,IQR$ and $Q_3 + 1.5\,IQR$; points beyond are plotted as outliers.
Box position and median show skew, and it compares many groups side by side.
A histogram plots frequency of values in bins, so it shows the shape of one distribution, while a box plot summarises spread and outliers compactly.

Example. Data 2, 4, 5, 7, 8, 9, 10, 12, 30: median 8, $Q_1=4.5$, $Q_3=11$, $IQR=6.5$, upper fence $=20.75$, so 30 is an outlier.

Answer frame. Define visualization and its importance; draw a box plot and a histogram; explain both; close by comparing use.

Asked: [7 marks] (Jun 2023) What is data visualization? Discuss box plots and Histograms.

Histograms

Definition. A histogram is a bar chart of the frequency of numeric values grouped into continuous bins. Bar heights show how many observations fall in each bin, so the shape of the distribution is visible.

Key points.

Bars touch because the bins are continuous, unlike a bar graph of categories.
Bin width is $\approx \frac{\text{range}}{\text{number of bins}}$, and too few or too many bins hide the shape.
It shows symmetry, skew, modes and gaps.

Scatterplots

Definition. A scatter plot draws each observation as a point at (x, y) for two numeric variables. It shows the relationship between the variables, such as positive, negative or no correlation.

Key points.

An upward cloud means positive correlation and a downward cloud means negative correlation.
It reveals clusters and outliers, and a fitted line shows the trend.
Colour or size can add a third variable.

Features Map Visualization

Definition. A feature map is the output of one filter of a convolutional neural network layer applied to an image, and visualizing it shows what the network detects. Feature map visualization displays these activations as images to interpret what each layer has learned.

Key points.

Steps: feed an input image, take the activations of a chosen convolution layer, and plot each channel as a grayscale or heat image.
Early layers respond to edges and textures, and deeper layers respond to shapes and object parts.
Bright regions mean strong activation, showing where the filter found its pattern.
Use case: checking whether a cat classifier looks at the cat rather than the background, which helps debug and explain the model.

Answer frame. Define visualization, then feature map, steps, interpretation, use case.

Asked: [7 marks] (Dec 2024) What is data visualization? Discuss feature map visualization technique.

t-SNE

Definition. t-distributed Stochastic Neighbour Embedding (t-SNE) is a non-linear technique that reduces high-dimensional data to 2 or 3 dimensions for visualization, keeping similar points close. It converts distances into probabilities and minimises the KL divergence between the high- and low-dimensional ones.

Steps.

Step 1: Compute pairwise similarities of points in high dimensions as Gaussian probabilities P.
Step 2: Place points randomly in 2D and compute similarities Q using a Student t-distribution.
Step 3: Minimise the KL divergence KL(P||Q) by gradient descent so that Q matches P.
Step 4: Plot the 2D points.

Key points.

Dimensionality reduction is needed because humans cannot see beyond 3 dimensions.
The heavy-tailed t-distribution avoids crowding of points.
Example: word embeddings of text or image features of digits give visible clusters of similar topics or classes.
Clusters mean similar items, but distances between clusters and cluster sizes are not reliable, and perplexity must be tuned.

Asked: [7 marks] (Jun 2025) Explain how t-SNE is used to visualize high-dimensional data with an unstructured data example.

Overview of Advanced Excel: Introduction

Definition. Advanced Excel means the analysis features of a spreadsheet beyond basic entry. It includes data validation, charts, pivot tables, scenario manager, protection and macros.

Key points.

It automates reporting and analysis of business data.
Functions, pivot tables and what-if tools reduce manual work.
Cell references and formulas keep results updated when data changes.

Data Validation

Definition. Data validation is an Excel feature that restricts what can be entered in a cell, so that only correct data is accepted. It improves data accuracy and reduces errors at the point of entry.

Steps. Select cells, then Data tab, Data Validation, choose the Settings type, set Input Message and Error Alert, click OK.

Key points.

Validation criteria include whole number, decimal, date, time and text length, each with limits such as between 1 and 100.
A list creates a drop-down of allowed values, for example Male, Female, so typing mistakes cannot occur.
A custom rule uses a formula, such as =COUNTIF($A:$A,A2)=1 to stop duplicates.
An input message shows a hint when the cell is selected.
An error alert (Stop, Warning or Information) appears when invalid data is entered and its style decides whether entry is blocked.
Circle Invalid Data marks existing wrong entries with red circles.
Example: marks cell allows whole numbers 0 to 100, and 105 shows the Stop alert.

Answer frame. Define validation and its role in accuracy; then techniques 1-6 with the steps; close with the marks example.

Asked: [8 marks] (Jun 2024, Jun 2025) Explain the few data validation techniques in Advance Excel / data validation in Excel and its role in improving accuracy and reducing errors.

Introduction to Charts

Definition. A chart is a graphical display of worksheet data, made using Insert, Charts. Visualization techniques are these chart types plus heatmaps and dashboards.

Key points.

Column and bar charts compare categories, for example sales by region.
Line charts show trends over time, for example monthly revenue.
Pie charts show parts of a whole, for example market share.
Scatter charts show the relationship of two numeric variables.
Heatmaps use colour intensity for values in a grid, and dashboards combine several charts and are used to monitor.
Choose by purpose: compare, trend, share or relationship.

Asked: [6 marks] (Dec 2024) List and explain the various charts in Excel. Asked: [7 marks] (Jun 2026) Explain different Data Visualization techniques in detail.

Pivot Table

Definition. A pivot table is an Excel tool that summarises large data by dragging fields into Rows, Columns, Values and Filters. A pivot chart is the chart linked to it.

Key points.

It aggregates with sum, count or average without formulas.
Fields can be rearranged instantly, and refresh updates it.
Example: sales data summarised as total sales by region and product, shown as a pivot chart.
In business reporting it gives fast, interactive summaries for decisions.

Asked: [7 marks] (Jun 2026) Discuss the importance of Pivot Tables and Charts in business reporting.

Scenario Manager

Definition. Scenario Manager is a What-If Analysis tool that stores sets of input values as named scenarios. It lets you compare results of best, worst and expected cases.

Key points.

Found at Data, What-If Analysis, Scenario Manager.
Each scenario changes chosen cells.
A Scenario Summary report compares outcomes.

Protecting Data

Definition. Protection prevents unwanted changes to a worksheet or workbook. Review, Protect Sheet locks cells with an optional password.

Key points.

Cells are locked by default, and locking works only once the sheet is protected.
Protect Workbook stops adding or deleting sheets.
File, Info, Encrypt with Password stops opening.

Excel Minor

Definition. Excel add-ins such as the Data Analysis ToolPak and Solver extend analysis. They are enabled from File, Options, Add-ins.

Key points.

ToolPak gives descriptive statistics, regression and histogram.
Solver finds optimal values under constraints.
Both need to be enabled first.

Introduction to Macros

Definition. A macro is a recorded sequence of actions or VBA code that Excel replays to automate repetitive tasks. A macro is recorded automation stored as VBA code.

Steps.

Step 1: Enable the Developer tab (File, Options, Customize Ribbon).
Step 2: Click Record Macro, give a name and shortcut key, and click OK.
Step 3: Perform the actions, such as formatting a header row.
Step 4: Click Stop Recording.
Step 5: Run it with the shortcut or Macros, Run; save as .xlsm.

Key points.

Macros save time and reduce errors in repeated work.
They are written in VBA, and can be edited in the VBA editor.
Example: a macro that bolds, colours and centres a header row in one keystroke.
Benefits are speed and consistency; limitations are security risk from malicious macros, and recorded macros cannot handle complex logic.

Answer frame. Define macro and purpose; steps 1-5 as a numbered list; example; close with benefits and limitations.

Asked: [7 marks] (Jun 2023, Jun 2024) Give an overview of Macros with suitable example / What is Macro? Write down the steps to create a macro in Excel.

Last-minute revision

Unstructured data has no fixed schema and is about 80 percent of data.
Four analytics: descriptive, diagnostic, predictive, prescriptive.
Box plot: five numbers, $IQR=Q_3-Q_1$, outlier beyond $1.5\,IQR$.
Histogram shows frequency in bins; scatter plot shows relationships.
Feature map is the activation output of a CNN filter.
t-SNE minimises KL divergence between P and Q.
Data validation types: number, list, date, text length, custom.
Alerts: Stop, Warning, Information.
Pivot table fields: Rows, Columns, Values, Filters.
Macros are recorded VBA, saved as .xlsm.

Memory hooks

DDPP: Describe, Diagnose, Predict, Prescribe.
Box plot: "min, Q1, med, Q3, max".
Validation: "Stop, Warn, Inform".
Macro: Record, Do, Stop, Run.

Coverage checklist

Importance of Unstructured Data: Q3.
Unstructured Data Analytics: Descriptive, diagnostic, predictive and prescriptive data Analytics based on Case study: Q5, Q11.
Data Visualization: boxplots: Q2.
histograms: Q2.
scatterplots: covered.
features map visualization: Q6.
t-SNE: Q8.
Overview of Advance Excel- Introduction: covered.
Data validation: Q1.
Introduction to charts: Q9, Q10.
pivot table: Q7.
Scenario manager: covered.
Protecting data: covered.
Excel minor: covered.
Introduction to macros: Q4.

Account

DATA SCIENCE (AD-404) - Unit 2 Short Notes

How unit 2 is examined

Importance of Unstructured Data

Unstructured Data Analytics: Descriptive, Diagnostic, Predictive and Prescriptive

Data Visualization: Boxplots

Histograms

Scatterplots

Features Map Visualization

t-SNE

Overview of Advanced Excel: Introduction

Data Validation

Introduction to Charts

Pivot Table

Scenario Manager

Protecting Data

Excel Minor

Introduction to Macros

Last-minute revision

Memory hooks

Coverage checklist