AI Forecast Charts: Predicting Tomorrow from Yesterday's Excel Data

Forecast charts project your historical data into the future so you can plan instead of guess. This tutorial covers Excel's built-in Forecast tools, plus how Copilot and ChatGPT extend them.

Table of Contents

• What Is a Forecast Chart
• Excel Forecast Sheet (Built-In)
• Forecasting with Copilot
• Going Further with ChatGPT
• Example: Python Forecast via ChatGPT
• Best Practices

What Is a Forecast Chart

A forecast chart plots your past values and extends a predicted line (with a confidence band) into future periods. It answers "if nothing changes, where are we headed?"

Excel does this with Forecast Sheet, which uses exponential smoothing — a method that weights recent data more heavily than old data.

Excel Forecast Sheet (Built-In)

1. Select your date column and value column (a table works best).
2. Go to Data → Forecast → Forecast Sheet.
3. Pick a line or column chart and set the Forecast End date.
4. Click Create.

Excel adds a new sheet with historical and forecast lines plus a shaded confidence interval.

Forecasting with Copilot

With a time-series table open, ask:

Forecast next 3 months of Sales and show it as a line chart
with a confidence range.

Copilot builds the forecast using the same engine as Forecast Sheet but in conversational form, which is handy when you want it on your current sheet.

Going Further with ChatGPT

Excel's forecast is one method. ChatGPT can suggest others (linear regression, moving average, seasonal models) and write the code to run them. This matters when your data has strong seasonality Excel smooths over.

Example: Python Forecast via ChatGPT

Ask ChatGPT for a seasonal forecast and it may return:

import pandas as pd
from statsmodels.tsa.holtwinters import ExponentialSmoothing

# df has columns: 'Date', 'Sales'
model = ExponentialSmoothing(
    df['Sales'], trend='add', seasonal='add', seasonal_periods=12
)
fit = model.fit()
forecast = fit.forecast(3)
print(forecast)

You run this in a notebook or Python tool, then bring the numbers back into Excel. Use this when Excel's single-line forecast feels too simple for your pattern.

Best Practices

• Forecast only with a clean, regularly spaced time series.
• Keep forecasts short — accuracy drops fast beyond a few periods.
• Show the confidence band so viewers see the uncertainty.