SeriesScout: Time Series

About this app

SeriesScout helps you explore patterns, compare forecasting methods, and understand what a sequence can—and cannot—predict. Enter a short number sequence for a quick next-value experiment, import a longer CSV dataset, or begin with one of the included real-world samples. SeriesScout compares simple baselines, statistical methods, and machine-learning models using validation on earlier observations. TEST FORECASTS HONESTLY - Reserve the latest observations as an untouched holdout set - Select models with rolling one-step validation on earlier data - Compare forecasts against actual holdout values - Review MAE, RMSE, relative error, and baseline improvement - See when a simple baseline performs better than a complex model COMPARE A BROAD MODEL LIBRARY Explore linear and quadratic trends, autoregressive models, seasonal methods, damped trend, Holt-Winters, logistic growth, power-law behavior, nearest-neighbor analogs, random forest, volatility models, and essential baselines. SeriesScout ranks applicable models using out-of-sample evidence rather than training fit alone. EXPLORE SERIES PROPERTIES Build a descriptive fingerprint of your data with measurements for: - Autocorrelation and short-range dependence - Candidate cycles and seasonality - Permutation entropy and local complexity - Volatility clustering and changing variance - Skewness, heavy tails, and extreme changes - Level shifts and possible regime changes WORK WITH YOUR OWN DATA - Import CSV files with a time or index column and a numeric value column - Choose how many recent observations to hold out - Optimize for exact values or directional classes with custom thresholds - Save imported datasets and experiment results locally LEARN FROM INCLUDED SAMPLES Explore weather, retail activity, atmospheric measurements, daily market movement, electricity demand, unemployment claims, sea ice, river flow, and earthquake counts. These samples include trends, cycles, shocks, noise, and changing volatility. Analysis runs on your device. SeriesScout is designed for exploration, education, and model comparison. Forecasts and diagnostics are estimates, not guarantees, and should not be treated as financial or professional advice.

Momentum comes from crawler snapshots for the tracked storefront. “Viral growth signal” means unusually fast recent rating gains; it is an indicator, not download data.

1. First observed Sep 29, 2026 · 9:02 AM $4.99