Forecast Accuracy Measures Screencast

Loading...

From the course by Rutgers the State University of New Jersey

Supply Chain Planning

551 ratings

Explore Related Videos

At Coursera, you will find the best lectures in the world. Here are some of our personalized recommendations for you

Rutgers the State University of New Jersey

Supply Chain Planning

551 ratings

Course 3 of 5 in the Specialization Supply Chain Management

Have you ever wondered how companies know how much to produce in advance so that they do not make too much or too little? Matching supply and demand requires planning. This course introduces you to the exciting area of supply chain planning. Part of a broader specialization on Supply Chain Management, you will master different forecasting techniques, essential for building a Sales and Operations Plan. At the completion of this course you will have the tools and techniques to analyze demand data, construct different forecasting techniques, and choose the most suitable one for projecting future demand.

From the lesson

Forecast Accuracy and Moving Average

In this module, learners will cover two more sophisticated forecasting methods, the moving average and exponential smoothing.

Forecast Accuracy Measures3:48

Forecast Accuracy Measures Screencast4:49

Meet the Instructors

Rudolf Leuschner, Ph.D.
Assistant Professor
Department of Supply Chain Management

Forecast Accuracy Screencast.

So now, what I would like to show you is how to calculate the three forecast

accuracy measures that we discussed in the lecture in a spreadsheet.

So, we are going to start with the Mean Error.

In order to calculate the Mean Error, you need to calculate the error for

each period first, which then you can average together.

So, we get the error by taking

our demand minus our forecast and

there we have it in period three demands

24 forecast is 22 our error is 2.

Now as we copy this down, what we see is

we have positive and negative errors.

We over and underforecast and that's okay.

That's what we expect from the error.

To calculate the mean error, then what we

do is we go over here where I've created a spot for

it and we take the average of all of these

errors that we have, then we have it the mean

error in this example is -0.71.

Now, let's look at the absolute percent error.

So, the absolute percent error needs two parts.

One is we need to get an absolute error and

then we need to convert it into percent.

So, we get the absolute error by taking the absolute value.

We already calculated error, so I'll keep using that and

then we need to divide it by our demand to get it into a percent form.

So, the absolute value of the error that we

previously calculated divided by demand and

then all I need to do is copy this down over here.

And now,

I go in to this cell that I prepared here from our mean absolute percent error.

All I need to do is here.

Average all of these values together and here's our result.

On average, we have an error of 13.58%.

So every time we create a forecast, we're about almost 14% off.

We don't know in which direction on average, we're off.

But in general, this is how far we're off on average.

So now, we are looking at the squared error and

the squared error simply the squared error value.

We raise it to the second power.

And all I need to do is drag this formula down,

captivated overall of these values and then I have another

spot where I see it's the average of all the values in the column G.

And here we go.

Our Mean Squared Error, MSE is 8.86.

Now 8.86 in general, we cannot really make a lot of statements on,

but on of the key features of the Mean Squared Error is that larger errors.

So for example, this forecast error of -7,

which is an over forecast of 7, now becomes 49.

So in an average, we give it much more than weight than

a small error, such as in period 5, 6 and 10,

which only has a value of 1 and 1 times 1 stays 1.

So what we see here with a Mean Squared Error is an amplification of large errors,

which is the ones that we really care about when we forecast.

We can deal with small errors, but large errors are the ones that make it

much more difficult to perform all of the steps that are required later on.

So here, we have the three accuracy measures.

The Mean Error, the Mean Absolute Percent Error and the mean Squared Error.

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.

Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.

© 2019 Coursera Inc. All rights reserved.

Download on the App Store

Get it on Google Play