Raw series (back row)
After transform (front row)
Negative value
⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.
Most real time series — sales, sensor readings, prices — drift: a trend pushes the mean up or down, or a random-walk component means past shocks never fade. Forecasting models like ARIMA assume the opposite, so the first real step of any time-series pipeline is testing for and removing that drift. This simulator lets you build a synthetic series from three tunable ingredients — deterministic trend, unit-root random walk, and a seasonal cycle — renders it as a 3D bar series, and lets you apply seasonal and regular differencing to it live. A real ordinary-least-squares Augmented Dickey-Fuller regression runs on whatever transform is currently active, so you can watch its t-statistic cross the classic −2.86 stationarity threshold as you difference the drift away.