Overview
What it is
The Micro AI Trend Forecaster is an interactive web dashboard that lets users explore fashion trend data and view AI-generated forecasts. Users upload or select a dataset, filter by category and season, and the tool surfaces trend patterns, colour distributions, and a 6-month sales forecast using a linear regression model trained on the data.
It started from a simple observation: traditional trend forecasting is expensive, slow, and locked behind large agencies. This project is an attempt to make that kind of intelligence more accessible using data that's already out there.
6mo
Forecast horizon
Linear regression model projecting 6 months forward
3+
Datasets supported
Upload any fashion CSV or use built-in Zara dataset
€29.50
Average price tracked
Across category, colour, and seasonal filters
The Product
Dashboard overview
Main dashboard showing category distribution, colour breakdown, and insight cards for Zara Tops dataset
Process
Key decisions
01
Linear regression over a more complex model
I chose linear regression for the forecast because the dataset size didn't justify a more complex model and transparency matters here. A simple, interpretable model that shows its working is more useful to a non-technical user than a black-box neural network with slightly better accuracy.
02
CSV upload as the core interaction
Rather than hardcoding a dataset, I built it around CSV upload so the tool works with any fashion data, not just the Zara set I used for testing. This made the product feel genuinely useful rather than a proof of concept.
03
Visualisation first, data second
The dashboard leads with charts and insight cards, not raw tables. The goal was to make the output feel like a tool you'd actually use rather than a data dump , the numbers should tell a story at a glance.