Data Visualisation Dashboard
Global Temperature Visualisation
Turning 40 years of climate data into an interactive story
Python, Pandas, NumPy, Matplotlib, Seaborn, Plotly, Plotly Dash
An interactive dashboard that visualises 40 years of global temperature data across 1,000 cities, built with a 4-person team. Rather than just presenting the numbers, it tests three specific hypotheses about warming through linear regression, so viewers see the trend itself, not just the raw data behind it.
01 / CONTEXT
What needed to change.
The key decision was not to just plot the raw temperature data and call it done. Numbers alone rarely change minds, so the team built the dashboard around three specific, testable hypotheses, whether average temperatures were rising, whether seasonal extremes were narrowing, and whether winters specifically were warming, and let the actual data confirm or challenge each one through linear regression rather than assertion. That meant choosing interactivity deliberately: a year slider and hover-based detail so a non-technical viewer, not just the team, could scrub through 40 years of data themselves and see the trend emerge, rather than being told it exists.
02 / WHAT I BUILT
A system built around the real constraints.
The dashboard presents 40 years of city-level temperature data through an interactive map with a year slider, letting users scrub through time and watch warming patterns emerge across 1,000 cities worldwide. A seasonal view isolates winter and summer trends to test whether extremes are narrowing, while hover-based detail surfaces city, country, and temperature values on demand. The data pipeline cleans and transposes the raw dataset with pandas and NumPy, then applies linear regression to validate each hypothesis before it's visualised. The whole service is containerised with Docker and deployed to a cloud VM, so the dashboard lives at a public link rather than a local script. The hypothesis charts below were built by teammate Alexandr, using the cleaned dataset and regression logic from our shared pipeline.
03 / WHERE IT STANDS
What works now, and what’s still being tuned.
The dashboard is finished and deployable via Dockerfile, with the team already running it live on a cloud VM rather than just locally. It remains open to extension rather than closed off: new hypotheses or city-level breakdowns could be added without restructuring the core pipeline, and the team has discussed continuing it past the original assignment.