Calculator Input
Example Data Table
This sample quarterly demand series works well for testing smoothing, trend, and seasonal settings.
| Period | Observed Value | Comment |
|---|---|---|
| Q1 | 120 | Starting demand baseline |
| Q2 | 128 | Moderate seasonal lift |
| Q3 | 133 | Steady underlying growth |
| Q4 | 145 | Strong late-season demand |
| Q5 | 138 | Next cycle begins |
| Q6 | 146 | Seasonal uplift continues |
| Q7 | 152 | Trend still rising |
| Q8 | 166 | Peak seasonal quarter |
| Q9 | 158 | Higher base than first cycle |
| Q10 | 169 | Demand expands again |
| Q11 | 176 | Healthy growth remains |
| Q12 | 188 | Latest peak observation |
Formula Used
Simple exponential smoothing: Lt = αYt + (1 − α)Lt−1. The next forecast equals the latest level estimate.
Holt linear trend: Lt = αYt + (1 − α)(Lt−1 + Tt−1) and Tt = β(Lt − Lt−1) + (1 − β)Tt−1. The m-step forecast is Ft+m = Lt + mTt.
Holt-Winters additive: best when seasonal swings stay roughly constant. Forecasts combine level, trend, and an additive seasonal index.
Holt-Winters multiplicative: best when seasonal swings rise with the series level. Forecasts use level, trend, and a multiplicative seasonal factor.
Error metrics: MAE is average absolute error, MSE is average squared error, RMSE is the square root of MSE, Bias is mean signed error, and MAPE is average absolute percentage error where actual values are nonzero.
How to Use This Calculator
- Paste your time series values into the data box using commas or new lines.
- Select the forecasting method that matches your data pattern.
- Set alpha, beta, and gamma values between 0 and 1.
- Enter the season length when your data repeats on a fixed cycle.
- Choose how many future periods you want to predict.
- Use automatic initialization for quick analysis, or test manual starting values.
- Press Calculate Forecast to show the forecast summary and the full result table above the form.
- Download the results as CSV or PDF after calculation.
Frequently Asked Questions
1. When should I use simple exponential smoothing?
Use it when your data has no strong trend or repeating seasonality. It works best for stable series where recent observations matter more than older values.
2. When is Holt linear trend better?
Choose Holt when values move upward or downward over time but do not show a strong seasonal cycle. It adds a separate trend estimate to the smoothing process.
3. What is the difference between additive and multiplicative seasonality?
Additive seasonality fits series where seasonal changes stay about the same size. Multiplicative seasonality fits series where seasonal variation grows or shrinks with the overall level.
4. Why do seasonal methods need two full seasons?
The calculator estimates initial seasonal factors and a starting trend from repeated cycles. Two full seasons provide enough structure for stable initialization.
5. How do I choose alpha, beta, and gamma?
Higher values react faster to new data but can make forecasts noisy. Lower values smooth more heavily and are usually better for stable patterns.
6. What does a lower RMSE mean?
A lower RMSE means your fitted forecasts stay closer to the historical observations on average, with larger mistakes penalized more heavily than smaller ones.
7. Why is MAPE blank for some rows?
MAPE divides by the actual value. When an actual value is zero, percentage error becomes undefined, so the calculator leaves that percentage entry blank.
8. Can I use this for sales, traffic, or inventory?
Yes. Exponential smoothing works for many ordered time series such as demand, website sessions, call volumes, production counts, and stock usage.