lecture 9 Flashcards
3 laws of forecasting
- forecasts are always wrong
- principle = it is very difficult to forecast demand exactly –> but very costly to ignore the demand variability - shorter forecast horizon –> better forecasts
- principle = short term forecasts are more accurate than long term forecasts
- reason = long term forecasts are noisy because of the many uncertainties - aggregate forecast are more accurate
- principle = forecasts are usually more accurate for groups
- reason = high demand somewhere cancels out low demand somewhere else
coefficient of variation
- standard of deviation = the amount of variation or dispersion of a set of data values –> measures the absolute variability
- coefficient of variation = ratio of the sd to the mean (CV = o/h kebalik) –> measures the relative variability
quantitative vs qualitative forecasting methods
qualitative =
- relies on qualities/characteristics and judgement of experts
- useful in new situations
- nature = subjective and inconsistent
quantitative =
- relies on quantities/measured values and statistical models
- useful when sufficient historical data is available
- nature = objective and reproductive
qualitative forecasting methods
- personal insight =
- use a single person to use his experience to forecast the future
- weakness = highly subjective - panel consensus =
- more people better than few people
- relies on a group of expert sharing information and arriving at an expectation through consensus
- weakness = a highly influential expert can influence the other experts - the delphi method =
- a structured group of people is better than an unstructured group of people
- relies on a group of expert answering questionaires on multiple rounds and resulting on anonymous summary
- strength = an influential expert can’t influence the other experts
quantitative forecasting methods
- linear regression
- logistics regression
- choice models
- classification and regression trees (CART)
demand = trends + seasonality + shocks
revenue management definition
maximizing revenue by predicting customer behavior at micro-market level (customer segments, individuals) and optimizing product availability
compete with low cost competitors (fixed cost and marginal cost of airline)
- fixed cost = independent on number of passengers (e.g. pay of pilots, airport fees)
- marginal cost = dependent on number of passengers (e.g. more weight –> more fuel, food and drinks)
if an airline has a lot of empty chairs –> marginal cost is near zero
4 core concepts of revenue management
- using variable pricing to shift demand
- use higher price equals to shifts some demand to low-demand periods - use market based pricing rather than cost based pricing
- charging what customers are willing to pay - selling to segmented micro markets
- ex : cinemas have price for students and adults - save inventory for valuable customers
- airlines know that business customers usually books late
- they save customers for business people with higher pricing
overselling
- sell more of something than is available or known as overbooking
- undersell –> empty seats (airplane)
- sell exactly –> empty seats because of no-shows
- sell more –> seats might be filled exactly –> maximizing profit –> service capacity is parishable
- gain : additional revenue from additional passengers
- loss : cost of compensating bumped passengers
overbooking level
- number of seats/room to overbook
- if selling price increases –> overbooking level increases (more profit)
- if compensation cost increases –> overbooking level decreaes (more cost)
- if no-show rate increases –> overbooking level increases (more empty spaces)
customer no show
- definition : customers that already reserves but didn’t show up nor cancel the reservation
- enough no-shows –> everyone is happy, too few no-shows –> unlucky customers will get bumped
- when facing bumped customers, companies have to act ethically and treat the bumped customers gently. They also have to provide incentives and government regulation may be necessary to protect customer’s rights
