Instacart Market Basket Analysis
“Happiness is not in money, but in shopping.”
- Marilyn Monroe
On 17th May 2017 Kaggle launched its Instacart Market Basket Analysis competition.
Instacart is a US-based grocery order and delivery app. Users select grocery products through the app, and personal shoppers review and complete the in-store shopping and home delivery.
The challenge posed in this competition - use open-sourced, anonymised data on 3 million Instacart orders over time to predict which previously purchased grocery items will be in a user’s next basket.
Entrants had 3 months to analyse, model and submit their entries, making as many entries as desired over this period to validate and improve their score.
I recently took part in a modified version of the Instacart challenge.
My challenge?
- Tackle the Kaggle Instacart Market Basket Analysis competition and present on my analysis and findings
My time limit?
- One week
The nature of the challenge was purposefully time-constrained, the goal being not to achieve the best F-score, but instead use the challenge as a platform to
- Create and communicate a narrative around my approach to the data and the challenge
- Situate the data problem firmly within its business context
- Make explicit my thought processes and reasoning
- Communicate ideas for future work
Below is the deck I prepared and presented.
