Vehicle Sales — Depreciation Grid¶
A make × model-year bar field: bar height is median selling price, so every make's depreciation curve stands in relief; bar radius is sales volume.
Data: Kaggle used-vehicle auction dataset (~559,000 sale records).
Try it: download this example below (or the full examples set), then open Vehicle_Sales_Example/Depreciation_Grid/DepreciationGrid_gv_node.csv — or just drag its folder onto the GlyphViz window.
What it is¶
How fast does a car lose its value? The answer for a half-million real auction sales is a field of bars. Each bar is one make in one model year: 20 makes by 18 model years, 338 bars in all, standing on a wire grid. Look along any row and the tops of the bars trace that make's depreciation curve in relief, rising from the old cars on the left to the new ones on the right.
The data are Kaggle's used-vehicle auction records, about 559,000 sales. A half-million messy records have been boiled down to a few hundred glyphs, and every attribute of every glyph is a statistic, so market structure you were not looking for is just visible.
How to read it¶
| Axis or property | Meaning |
|---|---|
| X (columns) | model year, from 1998 on the left to 2015 on the right |
| Y (rows) | make, ordered cheap to luxury, away from the viewer: Volkswagen and Mazda at the front, then Chrysler, Dodge, Chevrolet and on, to Lexus and Mercedes-Benz at the back |
| Bar height | median selling price: the depreciation curve for that make |
| Bar radius | sales volume, on a square-root scale, so a fat bar is a car that sold in numbers |
| Bar colour | median odometer reading, from dark blue for low mileage to yellow for high |
Click a bar for its numbers, for example Volkswagen 1999 | 74 sold | median $1.0k | -6.4% vs MMR | 127k mi. MMR is the Manheim Market Report, the wholesale book value the sale is compared with. A make's whole row is labelled with its median price, for example Volkswagen (median $9.8k).
What to look for¶
- The curves slope up to the right. Along each row the bars climb as the model years get newer. That is depreciation, drawn: a car is worth less each year older it gets.
- Luxury rows stand taller. The tallest bars in the back rows (Mercedes-Benz, Lexus, Infiniti, BMW and Audi) reach roughly two to four times the height of the tallest bars at the front, which is the point of ordering the rows by price.
- Colour follows age. The oldest bars are yellow, because old cars have high mileage, and the newest are dark blue. The colour is a second view of the same axis, so a bar whose colour does not match its neighbours' is a make-and-year with unusual mileage.
- Radius is where the volume is. Look for the years and makes with the fattest bars: they make up most of the sales.
Things to try¶
- Compare two makes by orbiting to look straight along the X axis, so that two rows overlay.
- Read the oldest cars. The leftmost bars are the cheapest and the highest mileage; the tags give the exact median prices.
- See the same data as a galaxy. Market Galaxy aggregates the same sales by make and model, and Raw Auction Field shows the sales unaggregated.
How it is built¶
The scene is 378 nodes: a ground grid, a title, the 338 bars, and 38 marker spheres that label the 20 makes and 18 model years along the edges. It is a static scene with no animation, built by generate_vehicle_sales_example.py from the Kaggle CSV with the three Vehicle Sales scenes sharing one generator. Bars are cylinders, so the radius reads as a width rather than a flat disc.
Data: Kaggle used-vehicle auction dataset (about 559,000 sale records).