Week 6 | Session 1: Analytics in SCM — Introduction & Big Data (6 Vs)

Course: Supply Chain Digitization — Module 3: Analytics in SCM

Session Agenda

What is Analytics? — Definition & 4 Characteristics

Analytics in one sentence: “Use data to create models which lead to decisions that create value”

Define objective: Before any analytics, must have a clear goal. What decision am I trying to improve?
Collect data: Data is the foundation of all analytics. Without good data, no analytics is possible.
Clean & process data: Raw data often has discrepancies, missing values, and errors → must be cleaned before use
Build model: Statistical model | AI/ML model | Optimisation model — must serve the original objective
Test & validate model: Ensure model performs reliably on new data before deployment
Take decisions: Model provides an analytical basis for choosing among multiple possible decisions
Implement & create value: Decision is implemented in the organisation → must generate measurable value

4 Key Characteristics of Analytics

Data: The input — raw material of analytics
Model: The engine — statistical, AI/ML, or optimisation
Decision: The output — an analytically-supported choice among alternatives
Value creation: The purpose — analytics must ultimately benefit the organisation

Why Data Matters — Global Data Growth

Data is the essential fuel for analytics. Without sufficient, quality data — no model, no decision, no value. Trend: Global data volume has grown exponentially over the last 15 years

Global data created, captured, copied & consumed — 2010 to 2025

Year	Global Data Volume	Note
2010	1.2 trillion GB	Baseline — internet becoming mainstream
2015	15.5 trillion GB	~13× growth in 5 years — mobile internet + social media proliferation
2020	59 trillion GB	COVID accelerated digitization → more companies went digital → more data generated
2025 (projected)	175 trillion GB	~146× more than 2010 in 15 years. Exponential growth driven by IoT, e-commerce, AI, and social media. (Source: Statista)

COVID effect: Pandemic forced massive digitisation across industries → more digital processes = more data
Key implication: With exponential data growth comes the need to understand and handle Big Data — hence the 6 Vs framework

Big Data — The 6 Vs Framework

Purpose: The 6 Vs help characterise and understand the nature of big data in any context

Four main Vs: Volume | Variety | Velocity | Variability → the 4 pillars of big data
Two additional Vs: Value | Veracity → concern quality and usefulness of data

The 6 Vs of Big Data — definition and SCM examples

#	V	Definition	SCM Example(s)
1	Volume	Huge amount of data created, captured, copied, and consumed	E-commerce orders: 24/7, globally, millions of customers → search data, shortlist, checkout, address, payment method all captured simultaneously. Inventory count: ERP data vs. physical warehouse count — discrepancies generate reconciliation data continuously.
2	Variety	Multiple data types and sources — image, audio, video, text, numerical	Customer reviews: numerical (star rating) + categorical (good/bad) + text (written complaint) + image (photo of damaged item) + audio (voice search). Quality inspection: image analytics compares chip photos to benchmark. AGV in warehouse: automated guided vehicles capture video of aisles.
3	Velocity	Data generated at very high speed — real-time, every fraction of a second	GPS tracking of trucks: latitude + longitude captured every second. Driver eyeball monitoring: image of driver’s eyes captured every second → alarm triggered if no eye movement for 2–3 minutes.
4	Variability	Inconsistency within the same type of data — data values fluctuate significantly	Demand data for electronics: spikes during Diwali, falls sharply after. Not just seasonal — customer behaviour has made demand far more variable. FMCG beauty products: last year’s bestseller may sell poorly this year due to shifts in perception.
5	Value	Data must be useful — generate actionable insight or decision-making benefit	Not all data collected is useful. Data that does not lead to better decisions or organisational benefit is a cost, not an asset.
6	Veracity	Data must be accurate and come from reliable, trustworthy sources	Social media gives real-time news in seconds — but also generates misinformation. Garbage in → garbage out: if input data is wrong, the model output is also wrong. Data scientists must validate source reliability.

Deep Dive into the 6 Vs

V1 — Volume: Large Quantity of Data

Example 1: E-Commerce Order Data
- Customer places orders 24/7 from any location in the world
- Millions of customers globally → enormous volume generated every second
Example 2: Inventory Count in Warehouse
- Problem: ERP system shows 10 units of a SKU → physical count may show 9 or 11
- Fix: Periodic physical inventory count → reconcile physical stock with ERP → generates large data sets

V2 — Variety: Multiple Data Types & Sources

Data types: Image | Audio | Video | Text | Numerical/Categorical

Example 1: Customer Review Data (Numerical: Star rating, Categorical: Excellent/Poor, Text: description, Image: photo of damage, Audio: voice search)
Example 2: Image Analytics for Quality Inspection (Microchip circuits are photographed on production line → compared with benchmark)
Example 3: Video Analytics for Inventory Count (Drone fitted with video camera flies across all aisles → scans QR codes)

V3 — Velocity: Data Generated at High Speed in Real-Time

Example 1: GPS Tracking of Trucks
- GPS captures latitude + longitude every second → real-time visibility of fleet
Example 2: Driver Eyeball Monitoring
- Image of driver’s eyes captured every second → if no movement for 2–3 minutes → alarm triggered

V4 — Variability: Inconsistency Within Data

Definition: The same data type shows large swings in value — often unpredictably

Example 1: Demand Data — Electronics
- Sales very high during festivals → drops sharply right after
Example 2: FMCG Beauty Products
- A product that sold in huge volumes last year may sell poorly this year

Variety vs. Variability — key distinction

V	What it means	Analogy	SCM Example
Variety	Different TYPES of data — image, audio, video, text, numbers	Different fruits in a basket (apples, oranges, bananas)	Customer review has: numerical rating + text + image + audio search
Variability	Inconsistency WITHIN one type of data — values change unpredictably	Same type of fruit but varying in ripeness, size, sweetness	Demand for electronics: low in Jan, very high in Oct (Diwali), low again in Nov

V5 — Value: Data Must Be Useful

Definition: Data generated must translate into organisational benefit — better decisions, cost savings, or revenue growth
Key question to always ask: “What decision does this data enable — and does that decision create value?”

V6 — Veracity: Data Must Be Accurate & Trustworthy

Definition: Data must come from reliable sources and be free of errors, fabrications, or manipulation
Fundamental principle: “Garbage in → Garbage out”
Responsibility: Always validate data source reliability, check for anomalies, and cross-verify before using data

Practitioner Checklist — How to Monitor the 6 Vs

#	V	What to Check Before Using Data
1	Volume	Is the data set large enough to build a reliable model? Do I have sufficient historical data?
2	Variety	Am I capturing all relevant data types? Is text-only sufficient or do I also need image/video data?
3	Velocity	Does my system handle real-time data? Is there latency in data capture that would make the output stale?
4	Variability	How stable is this data over time? Do I need to account for seasonal spikes or social media-driven swings?
5	Value	Will the output of my model lead to a decision that creates measurable value? Is this data worth collecting?
6	Veracity	Where does this data come from? Is the source trustworthy? Has it been validated?

Session Summary

Analytics in one sentence: “Use data to create models which lead to decisions that create value”
4 analytics characteristics: Data | Model | Decision | Value creation
Data volume: 1.2 trillion GB (2010) → 175 trillion GB (2025 projected) — exponential growth driven by digitisation and COVID
Big data: Characterised by 6 Vs — Volume, Variety, Velocity, Variability (4 main pillars) + Value + Veracity
V1 Volume: E-commerce order data (24/7 global) | Inventory count data
V2 Variety: Customer reviews (text + image + audio) | Chip quality inspection (image analytics) | Drone inventory count (video analytics)
V3 Velocity: GPS truck tracking (per-second location) | Driver eyeball monitoring (real-time drowsiness detection)
V4 Variability: Electronics demand (Diwali spike) | FMCG beauty (social media-driven demand swings)
V5 Value: Data must enable a decision that creates organisational benefit. Data without value is a cost.
V6 Veracity: Data must be accurate + from trustworthy sources. Garbage in → Garbage out.
Next session: Types of analytics (Descriptive, Diagnostic, Predictive, Prescriptive) + applications in SCM