~/Introduction to Big Data and Data Analytics

Introduction — Why Big Data Matters

In today's digital age, Big Data is a game-changer. Understanding its types, characteristics, and analytics helps us manage vast information, discover patterns, and unlock new possibilities in fields as diverse as healthcare, finance, entertainment, and environmental science.

🔹 Key Concepts You'll Learn

1. Introduction to Big Data
2. Types of Big Data
3. Advantages and Disadvantages of Big Data
4. Characteristics of Big Data (6 Vs)
5. Big Data Analytics
6. Working on Big Data Analytics
7. Mining Data Streams
8. Future of Big Data Analytics

Prerequisites: Understanding of the concept of data and reasonable fluency in English.

1.1 What Is Big Data? — Small Data vs Big Data

Small Data refers to datasets easily comprehensible by people — accessible, informative, actionable. Example: a small store tracks daily sales to decide what to restock.

Big Data refers to extremely large and complex datasets that regular programs cannot handle. It comes from three main sources:

• Transactional data — online purchases, banking transactions.
• Machine data — sensor readings, IoT devices.
• Social data — social media posts, tweets, comments.

1.2 Types of Big Data

2.1 Advantages of Big Data — 5 Key Benefits

🎯 1. Enhanced Decision-Making

Big Data analytics empowers organisations to make data-driven decisions based on insights derived from large & diverse datasets.

⚙️ 2. Improved Efficiency & Productivity

Analysing vast data helps businesses identify inefficiencies, streamline processes and optimise resource allocation.

👥 3. Better Customer Insights

Deeper understanding of customer behaviour, preferences and needs — enabling personalised marketing and improved customer experiences.

🏆 4. Competitive Advantage

Uncovers market trends, identifies opportunities, helps stay ahead of competitors.

💡 5. Innovation & Growth

Facilitates new products, services and business models based on data-driven insights — driving growth and expansion.

2.2 Disadvantages of Big Data — 5 Challenges

🔐 1. Privacy and Security Concerns

Large volumes of data raise risks of unauthorised access, data breaches and misuse of personal information.

❓ 2. Data Quality Issues

Ensuring accuracy, reliability and completeness is hard — Big Data often contains unstructured and heterogeneous sources, leading to errors and biases.

🛠️ 3. Technical Complexity

Implementing and managing Big Data infrastructure and analytics tools requires specialised skills & expertise.

📜 4. Regulatory Compliance

Organisations must meet data-protection laws like GDPR (General Data Protection Regulation) and India's Digital Personal Data Protection Act, 2023. Non-compliance invites legal risks and penalties.

💰 5. Cost & Resource Intensiveness

Acquisition, storage, processing, analytics and hiring skilled staff are expensive — a burden on smaller organisations with limited budgets.

3.1 The 3V and 6V Frameworks

Characteristics of Big Data start with the 3Vs — Volume, Velocity, Variety — and extend to the 6Vs by adding Veracity, Variability, Value.

⏩ 1. Velocity

Speed at which data is generated, delivered and analysed. Today millions of people store information online, so the speed is enormous.

📦 2. Volume

The sheer amount of data generated every day. Data crossing gigabytes falls into the realm of Big Data — ranging from petabytes → terabytes → exabytes.

🗂️ 3. Variety

Big Data spans many formats — structured, unstructured, semi-structured and highly complex. Ranges from simple numbers to text, images, audio, video. Unstructured data is hard to store in RDBMS but often provides the most valuable insights.

✅ 4. Veracity

Concerns consistency, accuracy, quality and trustworthiness of data. Not all data holds value — data must be cleaned before storage or analysis.

💎 5. Value

The business value derived from the data — perhaps the most critical characteristic. Without valuable insights, the other Vs hold little significance.

🔀 6. Variability

The consistency of the data stream even under extreme unpredictability. Ensures meaningful data can be extracted despite fluctuating conditions.

3.2 Case Study — OnDemandDrama (OTT Platform)

🔹 3V Framework

• Volume — huge data from millions of users: watch history, ratings, searches, preferences.
• Velocity — data processed in real time to adjust recommendations and surface trending content.
• Variety — diverse data: user profiles (structured), watch lists, video content, reviews — a mix of structured / semi-structured / unstructured.

🔹 Additional 3 Vs (6V Framework)

• Veracity — filters out irrelevant or low-quality data (e.g., incomplete profiles) for accurate recommendations.
• Value — personalises experiences to drive engagement and retention.
• Variability — handles inconsistencies in data streams from changing user behaviour, trends, events or regions.

4.1 Data Analytics vs Big Data Analytics

Big-Data Analytics covers the methodologies, tools and practices for data collection, organisation, storage, statistical analysis — used in business to refine processes, enhance decision-making and improve performance.

4.2 4 Types of Big Data Analytics (Recap)

From Unit 2 — Descriptive · Diagnostic · Predictive · Prescriptive.

4.3 4 Global Trends Driving Big Data Analytics

1. Moore's Law — exponential growth of computing power enables handling & analysing massive datasets.
2. Mobile Computing — widespread smartphones give real-time access to vast data from anywhere.
3. Social Networking — Facebook, Foursquare, Pinterest generate massive user-generated content and interactions.
4. Cloud Computing — access hardware/software remotely on a pay-as-you-go basis, no heavy on-premises investment.

5.1 4 Steps of Big Data Analytics

📥 Step 1 · Gather Data

Collect structured and unstructured data from various sources — cloud storage, mobile apps, IoT sensors, social feeds, internal databases.

⚙️ Step 2 · Process Data

Once collected, data must be processed correctly — especially when large and unstructured. Two common options:

• Batch Processing — examines large blocks of data over time.
• Stream Processing — processes small batches at once, giving quicker insight and shorter delay.

🧹 Step 3 · Clean Data

Scrubbing all data improves quality and results. Correct formatting and elimination of duplicate or irrelevant entries are essential. Erroneous or missing data lead to inaccurate insights.

📈 Step 4 · Analyze Data

Once the data is ready, advanced analytics turns Big Data into Big Insights.

🔹 Popular Big Data Analytics Tools

• Tableau — data visualisation.
• Apache Hadoop — distributed file system + processing engine for massive datasets.
• Cassandra — NoSQL distributed database.
• MongoDB — NoSQL document database.
• SAS — Statistical Analysis System.

5.2 Big Data Analytics with Orange — Heart Disease Dataset

📂 Step 1 · Gather Data

Use the File widget to load Orange's built-in Heart Disease dataset. Features: age, gender, chest pain, rest_spb (resting blood pressure), cholesterol, rest_ecg, max_hr, etc. Target: diameter narrowing (1 = narrowed arteries / heart-disease risk, 0 = healthy).

🧪 Step 2 · Process Data — Normalization

Use the Preprocess widget to normalise features (scale numerical values to a specific range like 0–1 or −1 to +1). Connect a Data Table to verify that all numerical values are now between 0 and 1.

🧼 Step 3 · Clean Data — Impute Missing Values

The Impute widget handles missing values. Strategies:

• Average (mean)
• Most frequent (mode)
• Fixed value
• Random value

Verify with a Data Table that missing values are replaced.

📊 Step 4 · Analyze — Build & Test Model

• Drop the Logistic Regression (or Decision Tree / K-Means) widget on the canvas.
• Add Test & Score widget. Connect learner + processed data.
• Double-click Test & Score → choose Cross-Validation.
• Connect Predictions widget to view outputs.
• Optional: use Scatter Plot · Box Plot · Heat Map to visualise patterns.

5.3 Mining Data Streams

6.1 Future of Big Data Analytics — 3 Key Trends

⚡ 1. Real-Time Analytics

Data processed instantaneously for immediate insights — live customer-behaviour monitoring, supply-chain tracking, fraud detection.

🧠 2. Advanced Predictive Analytics

Predictive analytics will evolve with sophisticated machine-learning and AI algorithms, enabling organisations to forecast trends and behaviours with greater precision.

⚛️ 3. Quantum Computing

Promises unprecedented processing power — quantum computers will solve complex problems much faster than classical computers, revolutionising Big Data analytics.