Reference no: EM132172174
CASE STUDY : Forecasting and Demand Management
At the end of 2016, to set its hotel room rates for the following year, the Department of Hospitality at FIU needs to forecast the demand for its rooms in 2017. The department has gathered 13 years of data (2004 to 2016) reflecting weekly bookings at its hotel.
You, as the Senior Demand Planner, are asked to forecast the weekly demand for their hotel rooms in 2017 using at least two to three different forecasting methods (groups of three students should at least do three methods). You have complete freedom in choosing your forecasting methods as long as they are logical and sound.
There Excel files are attached:
- "Case Study 2018 1--Training Set" contains the data for 2004-2015. You use this data to build your model.
- "Case Study 2018 2--Test data" contains the data for 2016. You use this data to test your models to decide which model is the best and you recommend to be used to forecast the 2017 demand.
- You can use the file "Case Study 2018 3--Dmd Forecast" to forecast the demand for 2017 using all your methods, marking your preferred method.
Note that holiday weeks in each year are marked during which demand can be higher than the normal weeks.
By the end of phase 1, due on or before Sunday November 18, you should report the following:
- Detailed analysis of the methods used and the estimation process
- Estimated parameters and the relationships/equations for your models, if applicable
- The error metric(s) for each method for your training and test sets.
- Forecast the weekly demand for 2017
- If possible, compute the prediction intervals for the first 4 weeks in 2017.
- Based on the error metrics for 2016 and the shape of forecasted values for 2017, use your judgement to choose one of the forecasting methods as your preferred alternative.
On November 19, you will be given the weekly observed demand in 2017. You will need to
- Calculate the SSE and standard error for all of your forecasting methods.
- Compare the errors and highlight the metric(s) for your preferred forecasting method from phase 1
- Report if your preferred method produced the best forecast or not.