Automatic Timeseries Forecasting

Predict the future of any timeseries

Upload your data, pick a horizon, and let Krafet run state-of-the-art forecasting models automatically. Built for students and researchers.

Get Early Access See how it works
Scroll

Features

Everything you need to forecast, nothing you don't

No boilerplate. No config files. Just results you can trust and cite in your next paper.

Auto Model Selection

Krafet benchmarks multiple models on your data and picks the best one. No manual tuning required.

Results in Seconds

Upload a CSV and get forecasts, confidence intervals, and diagnostics back almost instantly.

Export Anywhere

Download forecasts as CSV, PNG charts, or LaTeX-ready tables for your thesis or report.

Multiple Models

ARIMA, Prophet, XGBoost, LSTM, and more. Every run compares several approaches side by side.

Accuracy Metrics

MAE, RMSE, MAPE, and more — clearly reported so you can evaluate and compare models.

Free for Students

Academic accounts get generous free usage. No credit card needed to get started.

How it works

Three steps to a forecast

From raw data to publishable results in under a minute.

1

Upload

Drag and drop a CSV with a date column and one or more value columns.

2

Configure

Pick your forecast horizon and any constraints. Or just use the smart defaults.

3

Forecast

Get predictions, confidence bands, model comparisons, and exportable results.

Historical Data Forecast
Best Model
Prophet
MAPE
3.2%
Horizon
4 steps

Under the hood

Models we run for you

Each forecast automatically benchmarks across these approaches and reports the winner.

ARIMAClassical
ProphetDecomposition
XGBoostGradient Boosting
LSTMDeep Learning
ETSExponential Smoothing
ThetaDecomposition
N-BEATSNeural
EnsembleCombined