HYPERION Data Analysis Engine

LPG Demand Forecasting

Monthly forecasts with seasonal and trend decomposition using XGBoost

Executive Summary

Demand Intelligence

HYPERION LPG Demand Forecasting converts monthly national consumption data into accurate, seasonality‑aware forecasts using XGBoost, capturing long‑term upward trend and predictable annual cycles.

XGBoost

Model Type

Monthly

Cadence

Seasonal

Features

Confidence

Intervals

Intention & Objectives

Mission Critical

Deliver defensible monthly forecasts that resolve trend and seasonal structure for production, pricing, and stock decisions.
Reduce waste and stockouts by aligning supply with anticipated demand cycles.

Methodology

Pipeline Architecture

Feature engineering: Concatenate trend, seasonal Fourier terms, and lagged consumption. Modeling: XGBoost trained with time‑series CV.

Performance Metrics

Results Obtained

Strong correlation between actual and predicted consumption with robust holdout performance and consistent annual seasonality captured.