Quantium Data Analytics Job Simulation — Forage

---

Overview

This repository contains my completed work for the Quantium Data Analytics Job Simulation hosted on Forage.

The simulation placed me in the role of a data analyst on Quantium's retail analytics team, working with real-world transaction and customer behaviour data to deliver commercial insights and recommendations to a client Category Manager.

Analyst: Dinesh Navaneethan
Programme: Quantium Data Analytics | Forage
Completed: April 2025
Tools: Python · Pandas · NumPy · Matplotlib · Seaborn · Scipy · PowerPoint

---

Repository Structure

quantium-data-analytics-forage/
│
├── Task1/
│   └── QVI_Task1_Dinesh.ipynb        # Customer analytics & segmentation
│
├── Task2/
│   └── QVI_Task2_Dinesh.ipynb        # Uplift testing & trial assessment
│
├── Task3/
│   └── Quantium_Task3_Dinesh.pptx    # Client-ready presentation (Pyramid Principle)
│
├── outputs/
│   ├── plot_sales_by_segment.png
│   ├── plot_customers_by_segment.png
│   ├── plot_brand_affinity.png
│   ├── plot_price_per_unit.png
│   ├── plot_monthly_sales.png
│   ├── plot_pack_size.png
│   ├── plot_pretrial_alignment.png
│   ├── plot_trial_assessment_sales.png
│   ├── plot_trial_assessment_customers.png
│   └── plot_uplift_summary.png
│
└── README.md

---

Task 1 — Customer Analytics & Segmentation

Objective: Analyse transaction and customer behaviour data to understand chip purchasing trends across customer segments and deliver commercial recommendations.

What I did

• Loaded and performed high-level quality checks on QVI_transaction_data.xlsx and QVI_purchase_behaviour.csv
• Fixed date formatting (Excel serial integers → datetime), removed salsa products and outlier customers purchasing 200+ units
• Extracted PACK_SIZE and BRAND features from product name strings using regex
• Merged transaction and behaviour datasets and created a LIFESTAGE × PREMIUM_CUSTOMER segment column
• Analysed total sales, unique customers, units per transaction and average spend per segment
• Ran a two-sample t-test to confirm whether Mainstream Young Singles/Couples pay significantly more per unit than the rest of the population
• Computed brand and pack size affinity scores for the target segment

Key Findings

Insight	Finding
Top revenue segment	Budget – Older Families
Highest growth potential	Mainstream – Young Singles/Couples
Price premium confirmed	Mainstream Young Singles/Couples pay more per unit (p < 0.05)
Brand affinity	Over-index on Tyrrells vs general population
Pack size preference	150g–175g packs — consistent with impulse/on-the-go occasions

Sample Output

---

Task 2 — Experimentation & Uplift Testing

Objective: Assess whether the new trial store layout (stores 77, 86, 88) drove a statistically significant uplift in chip sales and customer numbers vs comparable control stores.

What I did

• Built monthly store-level metrics (total sales, unique customers, transactions) for the pre-trial (Jul 2018–Jan 2019) and trial (Feb–Apr 2019) periods
• Developed a control store matching function using two complementary similarity metrics:
  • Pearson correlation — tracks whether stores move together over time
  • Magnitude distance — measures whether absolute sales levels are similar
• Selected the highest-scoring control store for each trial store
• Scaled control store performance to the trial store's pre-trial baseline
• Used paired t-tests to determine whether trial period performance was statistically significantly different from control

Control Store Mapping

Trial Store	Control Store	Method
Store 77	Best matched via combined score	Pearson + Magnitude
Store 86	Best matched via combined score	Pearson + Magnitude
Store 88	Best matched via combined score	Pearson + Magnitude

Sample Output

---

Task 3 — Client Report (Pyramid Principle)

Objective: Compile findings from Tasks 1 and 2 into a structured, client-ready presentation for Julia (Category Manager), following Quantium's Pyramid Principle reporting standard.

Slide Structure

Slide	Content
1	Cover
2	Agenda
3	Executive Summary
4	Section break — Category
5	Category Overview
6	Customer Segmentation
7	Section break — Trial
8	Control Store Methodology
9	Trial Results — Sales Uplift
10	Trial Results — Customer Numbers
11	Recommendations
12	Next Steps
13	Closing

---

Skills Demonstrated

• Data cleaning & wrangling — handling missing values, outliers, date formatting, regex feature extraction
• Exploratory data analysis — segment profiling, distribution analysis, trend identification
• Statistical testing — two-sample t-tests, paired t-tests, p-value interpretation
• Control experiment design — Pearson correlation matching, magnitude distance scoring
• Data visualisation — Matplotlib, Seaborn, chart design for business audiences
• Business communication — Pyramid Principle, stakeholder-ready reporting, commercial recommendations

Certificate

Completed via the Quantium Data Analytics Job Simulation on Forage.

---

Built as part of an ongoing data analytics portfolio. Connect with me on LinkedIn.