Production-Ready MLOps Platform

Sales Forecasting Dashboard

End-to-end time-series forecasting using SARIMA, Prophet, XGBoost, and LSTM — with automatic model selection per US state.

▲ 100%

50

US States Covered

1,315

Avg RMSE

Root Mean Squared Error

▲ Low

8.6%

Avg MAPE

Mean Absolute % Error

▲ Strong fit

0.853

Avg R²

Coefficient of determination

1. Data Ingestion2. Data Validation3. Preprocessing4. Feature Engineering5. Model Training6. Model Evaluation7. Model Selection
Loading Sales Forecast...
Loading Best Model Distribution...

Best Model per State

Top 20 states shown

AlabamaProphet
AlaskaProphet
ArizonaProphet
ArkansasProphet
CaliforniaLSTM
ColoradoLSTM
ConnecticutLSTM
DelawareSARIMA
FloridaXGBoost
GeorgiaLSTM
HawaiiSARIMA
IdahoProphet
IllinoisProphet
IndianaProphet
IowaProphet
KansasLSTM
KentuckyLSTM
LouisianaSARIMA
MaineProphet
MarylandProphet
SARIMA
Prophet
XGBoost
LSTM
Loading Model Metrics...

Forecasting Models

SARIMAStatistical

Seasonal ARIMA with automatic order selection via pmdarima.

ProphetStatistical / ML

Facebook Prophet with yearly/weekly seasonality and US holidays.

XGBoostMachine Learning

Gradient-boosted trees using lag + rolling + calendar features; recursive forecasting.

LSTMDeep Learning

Multi-layer LSTM with dropout; sequence-to-one; recursive multi-step.