Using AI and ML for FP&A Forecasts

Hi,

Today, I want to double-click on how to use AI and ML to produce forecasts in FP&A.

With a real example.

That you can follow and use your own data on it afterwards.

Imagine you have to forecast revenue for your company.

And you need to leverage BIG DATA to do the forecast.

So basically a gigantic Excel file that crashes all the time.

But it does have the data you need.

Revenue, Expenses, Profits, Headcount, you name it!

So how to use AI for this?

Instead of using Excel to do your forecast, you will leverage Python and Prophet.

We have talked about Python, but if you don't know about it, read this first.

https://www.dhirubhai.net/posts/christianmartinezthefinancialfox_do-you-want-to-learn-python-for-fpa-but-activity-7205824139633246208-kAeH?utm_source=share&utm_medium=member_desktop

But what is Prophet.

Prophet is a powerful forecasting tool developed by Facebook.

It is designed to handle time series data, which is data that changes over time.

Prophet is particularly good at identifying patterns in your data and making accurate predictions. It is user-friendly and does not require deep expertise in statistics or machine learning to use effectively.

This below is an image of a revenue forecast created by prophet.

How Does Prophet Work?

Prophet breaks down your time series data into three main components:

1. Trend: The long-term increase or decrease in your data. For example, revenue might generally increase over several years.
2. Seasonality: Repeating patterns in your data. For example, you might see higher sales every December.
3. Holidays/Events: Special events that can affect your data. For example, a product launch or a marketing campaign might cause a spike in sales.

Implementing Prophet in Python

Here’s a step-by-step guide to use Prophet for forecasting your revenue:

1. Prepare Your Data: Collect and organize your historical revenue data. Ensure it has a date column and a revenue column. Clean the data to remove any inconsistencies or missing values.
2. Fit the Model: Use Prophet to fit a model to your historical data. Prophet will capture both the overall growth trend and any seasonal patterns. I have prepared this Google Colab notebook so you can just use this code.
3. Make Predictions: Once the model is fitted, you can use it to predict the revenue for the next year. Prophet will provide not only the forecasted revenue but also the range of uncertainty. And yes, you can download this forecast to Excel! (In the notebook I have added code for this)

Why Use Prophet?

By using Prophet, you can make more informed decisions in your financial planning.

This helps your company allocate resources effectively, anticipate future performance, and respond to potential risks and opportunities.

For example, if you have monthly revenue data for the past five years, you can use Prophet to fit a model to this data, capturing both the overall growth trend and seasonal spikes (like higher sales in December).

Once the model is fitted, you can predict the revenue for the next year, giving you an idea of expected sales and the range of uncertainty.

PS: In this example we have used Google Colab to run Python code but also Microsoft has announced the integration of Python within Excel!

So now it's time for us finance professionals to learn how to leverage this integration for data analytics, automation, and other ways to improve our work.

I partnered up with LinkedIn Learning and created this course about how to use Python in Excel!

You'll learn how to:

• Integrate Python with Excel for data manipulation, analysis, and reporting.
• Implement financial models and conduct data analyses.
• Automate repetitive tasks, enhancing efficiency.

Thanks for reading!

Christian Martinez