Interactive Time Series Studio

Rivu Basu Time Series Simulation Lab

Build, decompose, diagnose, and forecast time series in one browser-based lab. Mix trend, seasonality, cyclic behavior, and noise, then compare white noise, random walk, AR, MA, ARMA, and ARIMA behavior through live plots and interpretation.

Offline HTML ACF + PACF Forecasting CSV upload + download
Learning Focus
Components
Trend + Seasonal

See deterministic structure separated from noise and residual behavior.

Models
AR / MA / ARIMA

Explore dependence, shocks, differencing, and non-stationarity.

Diagnostics
ACF / PACF

Identify signatures like slow decay, lag cutoffs, and random structure.

Forecasting
Train vs Test

Compare fitted values, simple forecasts, and uncertainty bands.

Visualization

Time Series + Forecast Views

Trend Seasonal Residual / noise Forecast
A. Time Series Plot

Generated series

Hover values enabled
The full series combines deterministic components and stochastic dependence. Points: 0
B. Decomposition View

Trend, seasonal, residual

Toggle components on the left
Residuals show what remains after removing trend and seasonal structure. Components ready
C. ACF and PACF

Lag signatures

Lag hover enabled
AR models often show PACF cutoff behavior, while MA models often show ACF cutoff behavior. Lags: 20
D. Forecasting View

Training, test, forecast

Includes confidence band
Forecasts here use a lightweight teaching-oriented rule rather than a full optimization routine. Horizon: 16
Compare Mode

Model comparison summary

Current model in focus
Quiz Mode

Guess the model

Scenario ready

Loading quiz...

Use the clues in trend, stationarity, and correlation structure.
Quantity Value Meaning
Lag / Fit Metric Value Interpretation