liveBook · Manning

Point forecast

As explained in Time Series Forecasting Using Foundation Models

Deterministic forecasting is the technique of predicting one value for each future time step without representing forecast uncertainty.

A point forecast gives one predicted value for each future time step. Unlike probabilistic forecasting, it does not represent uncertainty as a range or distribution. section 7.0

A deterministic model directly outputs a future sequence. Models trained for this task can use losses such as mean absolute error or mean squared error; TimesFM uses MSE, which penalizes large errors more heavily. section 7.1

Point forecasts also arise when a probabilistic model is reduced to one representative value:

  • Use the median of sampled future values as the point forecast, limiting the influence of extreme samples. section 4.2.2
  • Extract the median at every horizon step when only point forecasts are needed from Moirai. section 10.2.6

Forecasting weekly store sales

TimesFM produces point forecasts for future values, while Lag-Llama and Chronos produce sampled forecasts from which medians and prediction intervals can be calculated. A plotted dashed median line represents the point forecast; shaded regions represent uncertainty and are separate from it. section 7.0 section 4.2.2

pip install timesfm

Install TimesFM to generate deterministic forecasts in Python. section 7.2

How Time Series Forecasting Using Foundation Models covers it

How deep it goes
Defines deterministic output, contrasts probabilistic distributions, and shows median extraction, but does not develop formal uncertainty methods.
Examples
Demonstrates TimesFM outputs, Lag-Llama and Chronos medians, and Moirai median extraction for weekly sales and blog visits.
What it assumes
Assumes basic time-series forecasting concepts and distinguishes single-value predictions from distributions without requiring advanced mathematics.
What you can do after
Apply point forecasts with TimesFM or extract medians from probabilistic forecast samples for each horizon step.
How it connects
Connects point forecasts with deterministic forecasting, probabilistic forecasting, prediction intervals, quantiles, median, MSE, and anomaly detection.
Cover of Time Series Forecasting Using Foundation Models
liveBookmanning.com
Read Time Series Forecasting Using Foundation Models on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Time Series Forecasting Using Foundation Models. The section links point to the parts of the book it draws on.