Riddhima Kapoor Fashion Technology @ NIFT Retail Merchandising  Consumer Behaviour Analytics

About Me

I'm Riddhima Kapoor, a Fashion Technology student at NIFT (BFT), with my foundation built in production merchandising. From there, my interest has moved toward the retail side of fashion — the business of it, where consumer behaviour and point-of-sale signal actually shape what gets bought and sold.

Along the way, I've picked up a habit of building things: tools and models that take scattered data and turn it into something people can actually act on.

That's the instinct I want to carry into retail merchandising and buying, where data meets the decisions that drive a brand forward.

Project Framework

raw data → derived metrics → aggregated summary → visual layer → decision.

1. → T&A Plans(Per style, per merchandiser)
2. → Data Consolidation(Style_Master + Generated_TA)
3. → Risk Scoring Engine(Weighted delay + buffer penalty, One row per style: score, category)
4. → Dashboard(KPIs, Top-10, hotspots)
5. → Merchandiser Action(Follow-up & escalation)

Integration with existing ERP:

• Built as a standalone Excel prototype for this phase
• Designed to eventually fold into CTA's existing ERP rather than replace it — sitting inside the Task Manager merchandisers already use daily

Future scope — AI/ML:

• Current model is rule-based and fully explainable by design (a manager can see exactly why a style scored what it did)
• Next step: use historical order and delay data to move toward predictive delay forecasting
• Goal: flag which styles are likely to slip — by buyer, process stage, or season — before the current milestone even shows signs of trouble, with today's rule-based score as the auditable baseline

Research Paper from NIFT Coursework

The Absence of a Standardized Framework for Measuring Sustainable Fashion Consumption: A Critical Review and Conceptual Proposal

The Problem:

Consumption — what people buy and why — drives most of a garment's lifetime environmental impact, yet it's the stage measurement handles worst. Industry tools (LCA, Higg Index) rigorously score products and factories but stop at the point of sale, and even Higg has been shown to misrank materials (polyester scoring "greener" than cotton) depending on weighting choices. Academic research fills the consumer gap but leans on self-reported surveys that poorly predict behavior (only ~30% of "ethical" intenders follow through when observed), leaving the field split and contradicting itself on basic questions like willingness to pay a sustainability premium.

The gap this creates:

• Studies contradict each other on basic questions (do consumers pay more for sustainability? how big is the attitude-behaviour gap?) because everyone's measuring something different

The proposed fix — ISCM, a 5-dimension framework:

1. Environmental Perception
2. Social & Ethical Perception
3. Cognitive Alignment (the novel piece — measures the gap between what people say and do, not just the attitude or the behaviour alone)
4. Economic Feasibility
5. Behavioural Engagement (the actual outcome — acquisition, usage, reuse, disposal)

Why it matters for Merchandising / Industry work:

• This is the same instinct behind my CTA Apparels dashboard — building a structured, comparable system where none existed, instead of relying on scattered, self-reported, or inconsistent inputs

Why clustering, specifically

You don't start with predefined customer "types" — you let the purchase data (frequency + spend) reveal natural groupings first, rather than assuming segments upfront. That's the justification for K-Means over just eyeballing the data or using arbitrary spend thresholds.

Data Processing

Why add KNN on top

Segmentation is only useful if it scales. K-Means gives you groups for customers you already have; KNN is what lets you slot every new customer into the right segment automatically, in real time, without re-running clustering each time.

That's the justification for combining unsupervised + supervised — one defines the strategy, the other operationalizes it. Small sample size — accuracy here is directional (proof of method), not a production-scale benchmark.

Model Evaluation

• Data split: 70% train / 30% test
• KNN applied with K = 5
• Performance evaluated using:
  • Confusion matrix
  • Accuracy

0.9666667Accuracy

How I Work

raw data → derived metrics → aggregated summary → visual layer → decision.

How I work:

• Find the actual bottleneck first — with CTA, it wasn't "no data," it was 15 disconnected files with no forward-looking view
• Prototype fast and cheap (Excel, R) before reaching for anything heavier — a rule-based, explainable model beats a black-box one until the simple version stops being enough
• Build for the person who'll use it daily — a merchandiser scanning a worklist in under a minute, not a dashboard that looks impressive in a review

Tools I use:

• Excel / VBA — for rapid prototyping, formula-driven systems, T&A and risk-scoring logic
• R (K-Means, KNN, ggplot2) — for segmentation, classification, and validating patterns in customer/behavioural data
• ERP & Task Manager systems — understanding how tools need to sit inside existing workflows, not replace them

CTA Apparels
Production Merchandising Internship

What I actually did

Order & Sample Management

• Tracked 25–30 styles simultaneously across the order cycle, from confirmation through factory handover
• Ran sampling end to end — Fit 1, Fit 2, Size Set, PP — interpreting buyer comments, routing alterations to the sampling floor, and chasing resubmissions until "OK for Production"
• Monitored T&A calendars against critical dates, flagging slippages before they became escalations

This is the exact bottleneck the Shipment Risk dashboard was built to solve.

Buyer Communication & Compliance

• Managed buyer communication on the Inditex portal — approvals, comments, follow-ups
• Designed a packaging reference ppt — fold method, polybag specs, main label, care label and hangtag placement, barcodes and carton markings — as a clearer, visual alternative to a standard excel sheet, shared with the buyer over the Inditex portal and appreciated for its clarity

Systems & Technical Exposure

• Worked in Visual Gems (ERP) for style, sample, and design entries; independently raised Material Requisitions for sample development
• Exposure to CAD, marker making, and fabric averages in costing

Cross-functional Exposure

• Prepared documentation for and attended a Pre-Production Meeting at a CTA unit — coordinating samples, tech sheets, swatches, PO, and recording action points
• Attended a Risk Analysis meeting for ZARA Lingerie styles

What's Next

I've seen how a product gets made. Now I want to own how it sells.

Working inside an export house taught me where product comes from — sourcing, T&A, buyer timelines, the operational reality behind every SKU. What I don't yet have full visibility into is what happens after: how that product performs at the point of sale, and why a consumer picks it up or walks past it.

My Graduation Project will focus on Retail Merchandising and Consumer Behaviour Analytics — building a structured way to connect POS-level signal (what's actually selling, to whom, how fast) with the buying and assortment decisions merchandisers make upstream. The instinct is the same one behind the CTA dashboard and my ISCM research: don't rely on scattered or self-reported inputs when a structured, comparable system can replace them.

What I will bring to the table:

• A merchandising-ops foundation — I understand what a buying, sizing, or promotion decision actually costs to execute upstream, not just how it looks on a dashboard
• Applied analytics skill (segmentation, classification)
• The instinct to build the structure myself when the data exists but isn't usable yet — same approach behind the Shipment Risk dashboard, now aimed at buying-side decisions: assortment planning, vendor coordination, and product performance tracking