Free data template

Simple forecast

A trend plus weekly pattern fitted to daily sales, the next 4 weeks forecast and a check against the last weeks.

Open in the Data LabSave and export

It opens with this template loaded and runs in your browser. You don't need an account to use it; a free account saves your work and exports it.

The job it does

For planning stock, staff or budget: a transparent baseline forecast before anyone reaches for a complex model. It comes with its own checks, so after you change it you can see at a glance that it still does the job.

How to use it with your own data

Load your daily numbers as a CSV called daily (columns date, value).

What's inside

This is the template exactly as the Data Lab opens it, read-only. Open it to run it, change it and export it.

  • 3 notebook cells
  • Data: the retail sample database (Mangrove Market: stores, customers, orders, payments)
The brief: Simple forecast

Planning stock, staff or a budget needs a forecast, and a transparent baseline comes first: a straight-line trend times a factor for each day of the week. The notebook checks itself on the last 4 weeks it was not fitted on (the average daily error, mape_pct), then forecasts the next 4 weeks with a chart.

Your own data: load a CSV file called daily with the columns date and value, one row per day, and Run.

Done when
  • The notebook runs top to bottom without errors
  • forecast has 28 rows
  • forecast.forecast has no empty values
  • A line chart plots forecast
Cell 1 · SQL into sales_daily
-- Revenue per day from June to November 2025 (delivered orders)
select o.ordered_at::date as date,
       (sum(i.qty * i.unit_price_cents) / 100.0)::float8 as value
from orders o
join order_items i on i.order_id = o.id
where o.status = 'delivered'
  and o.ordered_at >= '2025-06-01' and o.ordered_at < '2025-12-01'
group by 1
order by 1
Cell 2 · Python
import numpy as np
# Your own data: load a CSV file called "daily" (columns date, value) and it is used instead.
src = (daily if "daily" in globals() else sales_daily).copy()
src.columns = ["date", "value"]
src["date"] = pd.to_datetime(src["date"])
src = src.sort_values("date").reset_index(drop=True)
src["t"] = np.arange(len(src))
src["dow"] = src["date"].dt.dayofweek

def fit(df):
    """A straight-line trend, times a factor per day of the week."""
    slope, intercept = np.polyfit(df["t"], df["value"], 1)
    season = (df["value"] / (intercept + slope * df["t"])).groupby(df["dow"]).mean()
    return slope, intercept, season

def predict(model, t, dow):
    slope, intercept, season = model
    return (intercept + slope * np.asarray(t)) * season.reindex(np.asarray(dow)).to_numpy()

# check it first: fit on all but the last 4 weeks, forecast them, and measure the error
train, test = src.iloc[:-28], src.iloc[-28:]
check = predict(fit(train), test["t"], test["dow"])
mape_pct = round(float((abs(test["value"] - check) / test["value"]).mean() * 100), 2)
print(f"On the last 4 weeks the forecast was off by {mape_pct}% a day on average.")
Cell 3 · Python
# the next 4 weeks, from a model fitted on all the data
model = fit(src)
days = pd.date_range(src["date"].iloc[-1] + pd.Timedelta(days=1), periods=28)
t = np.arange(len(src), len(src) + 28)
forecast = pd.DataFrame({"date": days.strftime("%Y-%m-%d"), "forecast": predict(model, t, days.dayofweek).round(2)})

recent = src.tail(56)
plot = pd.concat([
    pd.DataFrame({"date": recent["date"].dt.strftime("%Y-%m-%d"), "actual": recent["value"].round(2), "forecast": np.nan}),
    forecast.assign(actual=np.nan),
], ignore_index=True)
chart(plot, x="date", y=["actual", "forecast"], kind="line", title="Revenue per day: the last 8 weeks and the next 4")
forecast
Open in the Data Lab

The lesson behind it

Learn why it works, with a podcast and a quiz for each lesson.

  • KPI dashboard notebook

    Weekly revenue, orders, average basket and active customers, with a chart per KPI.

  • A/B test readout

    Conversion per variant, the lift, a confidence interval and a plain-language verdict.

All data templates · Every template