Sign in or start a free trial to avail of this feature.
18. Limitations of the Model
Our model has provided some valuable insights to the client but it does have a number of limitations. The limitations of a model should always be highlighted to a client as well.
Limitations of the Zippy Airways Model (00:04)
A model is a simplified representation of reality, so it will always have some limitations. Some limitations for our Zippy Airways model are:
- It assumes a common overbooking policy for all flights. In reality, Zippy would probably want to apply different policies to different flights.
- It is backward looking, meaning it assumes the previous three months will be representative of the future. This may not be the case due to seasonality and other factors.
- It doesn’t use multivariable scenarios. This means that we can’t adjust multiple variables at once and see the effect on the model.
- It doesn’t account for unusual events, like flight cancellations due to weather.
In my definition of a model at the start of this course, I described models as simplified versions of reality. How simplified depends on time constraints, the quality of data and a number of other factors. The model in this course provides some valuable insights on an overbooking policy for Zippy Airways, but it also has some obvious limitations which should be highlighted to the client. Firstly, we assume a common overbooking policy for all flights. This would not be the case in reality where we could tweak the overbooking policy to suit each individual flight and maximize total additional profits. Next, the model is backward looking. We rely on the first three months of the year to inform us on demand for the remainder of the year. It would have been better if the company could have provided some alterations to this data that allowed for seasonality which can affect the airline industry. Thirdly, we didn't use multivariable scenarios. Our sensitivity tables allow us to alter two variables at the one time. In this case booking limit and conversion rate. Scenarios which we will explore in the next modeling course allow you to adjust many variables at the one time. So in our Zippy example, we could have run a scenario which had varied voucher costs, conversion rates, SlowMo costs and booking limits all at once. Scenarios are particularly useful in more complex models which may hold 10 to 15 different variables. The last limitation of our model that I'd like to mention is its inability to account for unusual events for example flight cancellations due to poor weather. In this particular case, when the company can alter the booking policy daily to respond to unusual events, the impact may not be substantial. But for certain decisions such as launching a new product, which cannot be undone, allowing for unusual events is critical. We'll learn how to do this using Monte Carlo simulations in a future modeling course.