time-series-forecasting-electricity-0001

Use Case and High-Level Description

This is a Time Series Forecasting model based on the Temporal Fusion Transformer and model trained on the Electricity dataset.

Specification

Metric

Value

GOps

0.40

MParams

2.26

Source framework

PyTorch*

Accuracy

Metric

Value

Normalized Quantile Loss (P50)

0.056

Normalized Quantile Loss (P90)

0.028

Normalized Quantile Loss described in Bryan Lim et al..

The quality metrics were calculated on the Electricity dataset (test split).

Input

name: timestamps shape: 1, 192, 5 format: B, T, N B - batch size. T - number of input timestamps. N - number of input features.

Output

name: quantiles shape: 1, 24, 3 format: B, T, Q B - batch size. T - number of output timestamps. Q - number of output quantiles (0.1, 0.5, 0.9).

Demo usage

The model can be used in the following demos provided by the Open Model Zoo to show its capabilities: