Portfolio / Case study 04

ForecastingPythonPlanning

Revenue forecasting model

A Python and Pandas model that projects revenue from sales history and shows a likely range rather than a single guess.

ClientForecasting project
IndustrySmall business planning
My roleAnalyst
ToolsPython, Pandas, Excel
Revenue forecasting model for Forecasting project

Project walkthrough

See how the work was put together

Revenue forecasting model in Python, project walkthrough
Revenue forecasting modelClick to play
finzain · project log

$ start project "Revenue forecasting model"

✓ Brief agreed with Forecasting project

✓ Cleaned and grouped historical sales by month

✓ Tested seasonal and trend-based forecasting methods

✓ Presented a base case with low and high ranges

→ Review with the client…

● Delivered and handed over

The project

Problem, approach and result

01

The challenge

Planning relied on last year plus a percentage. Seasonal swings and one-off months made that number unreliable.

02

What I did

  • Cleaned and grouped historical sales by month
  • Tested seasonal and trend-based forecasting methods
  • Presented a base case with low and high ranges
03

The result

Budget conversations started from a realistic range, with the seasonal months called out in advance.

Why last year plus a percentage falls short

Many small businesses plan by taking last year’s revenue and adding a percentage. It is quick, but it hides the shape of the year. A business with a strong season and a quiet season does not grow evenly, and one unusual month, such as a large one-off order, can push the whole estimate off course.

The aim of this project was not to predict the future perfectly, which no model can do. It was to give planning conversations a more honest starting point, one that shows how uncertain the number really is.

How the forecast is built

I cleaned the historical sales data and grouped it by month so that the seasonal pattern could be seen clearly. In Python and Pandas I tested seasonal and trend-based methods side by side and compared how each one would have performed on months that were already known. The steps were:

  • Clean the history and flag one-off months so they do not distort the trend
  • Group sales by month and look for seasonal swings
  • Test different forecasting methods against known months
  • Present a base case with a low and a high range instead of a single figure

Using a range in budget conversations

A single forecast number invites an argument about whether it is right. A range changes the discussion to how much risk the business is willing to plan for. The seasonal months were called out in advance, so the budget could allow for them rather than being surprised.

If you are budgeting from last year plus a percentage today, a simple first step is to chart your sales by month for the last two or three years. The pattern you see is usually the pattern your forecast needs to respect.

Gallery and links

Screens from the project

Figures are anonymised. Click an image to enlarge it.

Revenue forecasting model main screen
Main view
Revenue forecasting model detail screen
Detail
Revenue forecasting model report screen
Report

Let's connect

Want something like this for your business?

Book a free 30-minute call on Google Meet, or reach me directly.