Week 5 Lecture 1 - Forecasting Flashcards
What are the 4 steps in the forecasting process?
- Identify the purpose of the forecasting.
- Are we trying to forecast daily sales? Yearly sales? Growth of the market? - Series to be forecast
- Are we going to forecast one element of sales? One product? A series of products? A group of products? - Classification of series
- statistical forecast - Forecast adjustments
- The forecasting process is adjusted over time.
What is the ‘purpose’ of forecasting?
- Strategic Planning
- HR Planning
- Financial Planning
- Sales and Marketing Planning
- Operations planning
If the forecasting period is long, what happens to the interval of data points?
They get spaces apart more. It would be unrealistic to forecast every single day.
What is the downfall regarding forecasting?
All forecasts will be wrong, the aim of forecasting is to find a number/quantity that is close to the final outcome.
What three forms does forecasting take?
- Statistical (objective) methods - usually used more in the short term e.g. time-series, causal methods
- Judgmental (subjective) methods - a group of experts try to find a forecasting value. Useful for long term forecasting or when introducing a new product. E.g. individual judgment, group judgment.
- Guessing
What flows towards and away from the customer in a supply chain forecast?
Materials flow towards the customer
Information flows back away from the consumer to the supply chain
A high level of aggregation and a short time horizon is best forecast using what methods?
Time series
And
Causal methods
A low level of aggregation and a short term horizon is best forecast using what methods?
Time series
A high level of aggregation and a long horizon is best forecast using what methods?
Judgemental methods
A low level of aggregation and a long horizon are best forecast using what methods?
Causal
And
Judgmental
What four factors influence the use of ‘judgment’?
- No data (new product)
- Historical data is no longer relevant (product has been superseded)
- Different circumstances (new promotion, new discount level)
- An expectation that human judgment will be more accurate
What are two downsides of using the judgmental method?
They’re time consuming
They need employees to complete
In relation to judgment, cognitive biases come in three forms, what are they?
Optimistic bias
The forecast is too optimistic.
Confirmation bias
You already think something to be true so when you find 1 piece of supporting evidence, you accept this as absolute.
Overconfidence bias
This relates to confirmation bias but relates to people being overconfident without conducting the right amount of research.
What are the 3 main reasons for optimistic bias?
Confusion of targets and forecasts.
Over confidence
Competitive neglect
What is the ‘Delphi method’ and when is it used?
The ‘Delphi method’ involves trying to reach a general consensus through multiple rounds of questionnaires completed by a group of experts.
The questionnaire is distributed multiple times and the results aggregated and distributed to the group who then take the questionnaire again.
What 3 main ‘things’ can a time-series graph show?
- Trend
- Seasonality
- Irregularity
What is the definition of a trend in terms of a time-series graph?
A consistent change in the level of the time series
What is ‘Line Shift’ in relation to a time-series and what causes it?
Line shift is a sudden change (jump) in the level of demand.
A line shift can be a result of changes in conditions.
What is a ‘trended series’?
A trended series assumes that a trend will continue throughout the forecasting horizon.
How is a trend line identified in a trended series?
Trend line can be used in excel. It works by using ‘least squares’ linear regression to identify the line of best fit.
When is a trend line appropriate?
When the data is non-seasonal
What is meant by the term ‘Seasonality’?
Periodic changes occurring in a repeating pattern (e.g. daily, weekly, annually)
What is ‘not true’ in respect to seasonality?
Seasonality is not random
What is meant by the term ‘Cyclical Variations’ and how long is a cyclical term?
Variations that occur within one cycle (Economic cycle). Cycle lengths vary and therefore it is difficult to predict.
Cyclical trends typically occur over a year or more.
Irregular activity can also be referred to as …
Noise
What is meant by the term ‘Irregular Activity’?
Irregular activity is what is left over after accounting for trends, seasonality and cyclical variations.
Irregular activities represent the variations that we cannot define and as such cannot predict
What are the four steps to forecasting?
- Understand the data - Are we expecting variations? Draw a graph?
- Select the forecasting method - Should we pick one forecasting method over another because we are expecting variations?
- Apply the most appropriate method - We pick the method most appropriate for the data. How will this method be applied? Software? Manually calculated?
- Review subsequently - How well did the forecast go?
What is meant by the term ‘Naive forecasting’?
Using the exact value of the last periods data to compute the sales expected in the next period.
No adjustments or attempts to establish causal factors are made.
What is meant by ‘The Extrapolation method’?
The Extrapolation method involves using just two values, the growth/loss on the past periods results added to the last periods results.
Example:
2015: 50 sales
2016: 65 sales (15 increase)
2017: 65+15 = 80 sales
What is meant by the term ‘Central moving average’?
The central moving average involves using extrapolation but over a longer period of time. An average over a number of periods (2 or more) is used to compute an average which is added to the previous periods results.
When choosing the number of periods to be used in the calculation of the moving average , what two things must we consider?
What are we aiming to do?
- Identifying trends?
- Smooth out noise?
What time scales are we looking for?
- Averages using fewer periods will more accurately reflect short term trends.
- Averages using more periods will more accurately reflect long term trends.