Moving Average Forecast Calculator

Analyse time-series data, compare moving average methods, optimise windows, measure forecast accuracy, visualise trends, and export clear machine learning results instantly online with confidence.

Calculator Inputs

Use commas, spaces, semicolons, or new lines. Use NA for missing values.
One numeric column, or label and value columns. Maximum 2 MB.

Moving Average Method

Weights are automatically normalised.

Preprocessing


Validation and Optimisation

Supply future observations to evaluate the generated forecast directly.

Advanced Options

Formula Used

Simple Moving Average

SMAt = (xt-1 + xt-2 + ... + xt-n) / n

Weighted Moving Average

WMAt = Σ(wixt-i) / Σwi

Exponential Moving Average

EMAt = αxt + (1 − α)EMAt-1

Moving averages mainly estimate a local level. Trend and seasonal settings extend that behaviour.

How to Use

  1. Enter ordered observations or upload a valid CSV file.
  2. Select a moving-average method and suitable window size.
  3. Choose preprocessing, validation, and advanced forecasting options.
  4. Calculate, compare accuracy metrics, and inspect forecast intervals.
  5. Copy results or export the table, chart, and report.

Example Data

MonthObserved demandPurpose
January120Starting observation
February128Second observation
March125Completes a three-period window
April134First one-step forecast comparison

Frequently Asked Questions

What is a moving average forecast?

It predicts future values using averages of recent observations. The window controls responsiveness and smoothness.

How should I choose the window size?

Use smaller windows for fast changes. Use larger windows for smoother, slower patterns.

When should I use WMA?

Use WMA when recent observations should influence forecasts more strongly. Custom weights define that influence.

How is EMA different from SMA?

EMA applies exponentially declining weights. It usually reacts faster than an equal-weight SMA.

What does RMSE measure?

RMSE measures typical forecast error while penalising large misses more heavily. Lower values are better.

Why can MAPE be unavailable?

MAPE divides errors by actual values. Zero actual values make that percentage undefined.

Does a moving average model trends?

A basic moving average estimates level only. Enable trend adjustment or double moving averages for trend.

What is walk-forward validation?

It repeatedly forecasts the next observation using earlier data. This helps prevent future-data leakage.

Are forecast intervals guaranteed?

No. These intervals use historical residual error assumptions. Structural changes can reduce their reliability.

Related Calculators

Exponential Smoothing CalculatorForecast Error CalculatorTime-Series Train-Test Split CalculatorLag Feature CalculatorMean Absolute Scaled Error CalculatorWalk-Forward Validation Calculator

Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.