Forecasting Models PPT
khaled054
h3_1.pptx
TOP 10 Forecasting models
Meghan Woods
Marketing 188
Dr. William Rice
4:00- 5:50 pm T-TH Class
Row 2, Seat 1, Group 14
Econometric model
Description: These statistical models identify the relationships between various economic entities within a given study. Econometric models are often arranged under a certain economic theory and the forecast is built around that theory to support it. Economists often use this technique to determine future developments and identify what outcomes they may take in the market.
Advantages:
Only solution to “what if” scenarios
Research accompanied by economists input
Disadvantages:
Merely approximations to reality
Unknown parameter values
1
http://home.iitk.ac.in/~shalab/econometrics/Chapter1-Econometrics-IntroductionToEconometrics.pdf
Real world application: Econometric models are used by marketers and economists alike to forecast when making decisions in policy formation. Fitted models are often a real world representation of economic elements that policy makers must adjust when they see fit.
A set of equations represents the economic behavior occurring in a given market. --->
These results are graphed for forecasters to better interpret results that are then reviewed by economic analysts, and a decision is then reached on what actions to take or not take.
Econometric model
1
Econometric model
1
diffusion index
Description: Used often by economists and traders, this forecasting technique is a summarization of common tendencies that occur within a given data set. A statistical series is analyzed and interpreted by forecasters; if the series shows a greater number of rising data than declining, then the index number is above 50.
Advantages:
More participants likely to respond
Smaller mean-squared errors
Prompt results
Less data crunching
Confidentiality remains intact
Disadvantages:
Small changes cause big change in results
Changes not correlated in results
2
diffusion index
http://www.marketthoughts.com/z20050530.html
2
life cycle analysis
Description: Product life cycle analysis is a quantitative technique of forecasting. It revolves around patterns of past demand in data. This data encompasses these phases that are shown upon a curve model: introduction, growth, maturity, saturation, and decline. Phases of the life cycle help forecasters know when to best execute certain actions based upon similar products.
Pros:
Good for benchmarking performance
Stakeholder engagement tool
Maximize value
Reduce waste
Cons:
Not reliable predictor of true lifespan
False assumptions of life cycle
http://www.environmentalleader.com/2012/03/21/the-benefits-of-life-cycle-analysis/
3
life cycle analysis
Introduction:
small market size
expensive to implement
low sales
high researching & testing costs
Growth:
growth in sales & profit
increase in investment
economies of scale
Maturity:
maintain market share
product modifications and improvement
more efficient production
Decline:
product market shrinks
market becomes saturated
consumers switch brands
http://productlifecyclestages.com/
3
Trend projection
Description: A trend shows changes in the market over time. In this case, researchers use historical data to match against these trends and extend outcomes into the future.
Strengths:
Efficient
Fast to calculate
Easy to update/manipulate data
Weaknesses:
Unclear answers/distortion
Not reliable for long-term projections
http://smallbusiness.chron.com/pros-cons-trend-analysis-forecasting-58786.html
4
Trend projection
http://www.babylontoday.com/
Credibility of results lessens as time passes into the future, because projections are based off past results.
Best used only 4 years into the future, because after that point cyclical demand is identified and accuracy tends to fluctuate
Equation for trend projection:
Y= a + bx
4
Linear Regression
Description: Regression forecasting is used to determine relationships between dependent and independent variables. It serves its purpose when used in causal models- this is because regression analyzes multiple variables at once and measures the impact each one has.
https://neerajbhatia.files.wordpress.com/2010/10/linear_regression1.pdf
Strengths:
Identify correlations between variables in the marketplace
Lends scientific angle to support data
Can correct errors made by management, yield new valuable information
Weaknesses:
One sided view of linear relationships that could be incorrect
Data is sensitive to outliers
Data in the study must be independent
5
Linear Regression
Example of the company, AT&T using linear regression analysis to determine the relationship between buying times and prices of the product.
5
Linear Regression
http://mobiledevmemo.com/when-why-and-how-you-should-use-linear-regression/
When it is viable:
The relationship between the two variables (dependent & independent) must be linear
Data must be homoskedastic
Residuals must be distributed randomly
Residuals must be a normal distribution
5
The Delphi Method
Description: This interactive model is based on the results of respondents that answer a series of questionnaires in succeeding rounds. Their answers are then analyzed by experts and used to make forecasts. Although it is a useful method to get participants involved, they are not subjected to biased decision making and results are interpreted to be more accurate.
Strengths: Anonymity valued - Highly structured - Avoids unfair influences
Weaknesses: Costly - Takes long increments of time - Accuracy can vary
http://www.robertsevaluation.com.au/2013-09-12-04-14-43/delphi-technique
6
The Delphi Method
The Delphi method is best to use when planning on launching a new product to predict:
Success of the product in the market
Used to predict future sales
Forecast potential sales margins
6
The Delphi Method
Examples of Delphi Method Process
6
Moving Averages
Description: Used to identify trends in consumer purchasing, sales, etc. after previous periods have been recorded, this is makes moving averages a “lagging indicator”. Also this method is frequently used for predicting post-period trends, assuming consumer behavior will stay constant.
Pro: Smooths variation and seasonal variation among periods, isolating a trend- making it easier to identify and use for marketer analysis.
Con: Can be unreliable source of data. Once the data set is “cleaner”, outliers are removed and information can be less accurate.
http://www.shmula.com/forecasting-unweighted-and-weighted-moving-average-model/308/
7
Moving Averages
When we use them:
We need moving averages to recognize reversals and trends in the market. They determine the ability of a product or service to survive or succeed in specific areas of the market.
7
Moving Averages
http://zaielacademic.net/excel_vb_advanced/moving_averages.htm
7
Exponential smoothing
Description: This statistical method removes random variation, making more clear patterns within a given data set - hence the term “smoothing”. This is achieved by giving increasing weight to more recent, and therefore more relevant (in the eyes of the marketer) data.
Strengths: The process of smoothing is done mathematically, which means inexpensive and efficient in the eyes of marketing analysts.
Weaknesses: Does not accurately model seasonality or trend data.
http://www.inventoryops.com/articles/exponential_smoothing.htm
8
Exponential smoothing
https://webanalyticswlp.wordpress.com/2011/01/21/forecasting-methods-regression-vs-exponential-smoothing/
8
Panel Consensus
http://www.123helpme.com/forecasting-methods-view.asp?id=165096
Description: A panel makes use of the internal and external information within a supply chain by initiating interactive roles for each member. Historical data is added upon, according to new market trends to ensure accuracy.
Positives:
Supports supply-chain management decisions
Encourages equality in decision making process
Negatives:
Varying accuracy of “top performers”
Conflicting ideas between forecasters
9
Panel Consensus
How it works:
Experts are shown data from members within the supply chain and decide on which steps to take next by making intelligent decisions based on forecasts they’ve made.
Real life application:
In this case, the supply chain managers are trying to decide what steps to take for a farm crop corporation. Farm data, strategies and policies are reviewed in succession before action is taken by the actual farming company and its supply chain managers.
http://www.fao.org/docrep/006/ad237e/ad237e02.htm
9
Historical Analogy
Description: Historical analogy uses equivalence groups to make more accurate predictions than judgement and research alone. It does this by identifying a history of sales from similar products or situations to forecast future sales using trends.
Pros:
Good for testing products susceptible to change, or new products to the market
Cons:
Best used in specific situations only
The past isn’t always a good source of relevant information
http://wikieducator.org/User:Kavita11/Business_Forecasting
10
Historical Analogy
New product forecasts are projected using data analysis from similar or related product history
Testing to see if similar products thrived in the market based on historical data
10
the apricot problem: case study
Information: Apricots are highly popular in the central valley region. When they are in season, they are bought by consumers very quickly even though the first of the season products are highly priced.
Problem: The vast variety of apricots, weather and historical expectations are all factors that attribute to production issues. The problem for distributors trying to inform buyers on the East coast about when their crop will be available is unpredictability.
Failed solution: Crossbreeding has failed as a solution because the apricots aren’t produced to their standard.
the apricot problem: case study
New proposed solution:
Gather data from farmers about seasonality
Use data to interpret information from scientists about what the best way to cross-breed apricots and when
Get approval from panel of expert forecasters
Inform buyers on East Coast of new production schedule
Implement MRP system of distribution chain to save buyers money and time
Forecasting solution: Panel Consensus
Goal: To include everyone in supply-chain
the apricot problem: case study
Step 1: Gather data from farmers about seasonality
Climate requirements:
Grow in dry temperatures
Above 3000 sea level
Long cool winters & warm springs
Soil requirements:
Well drained, deep soils
Good organic matter
pH of 6.0-6.8
the apricot problem: case study
Step 2: Use data to interpret information from scientists about what the best way to cross-breed apricots and when
the apricot problem: case study
Step 3: Get approval from panel of expert forecasters
Checking with professional forecasters before making any rash decisions about purchasing and when to distribute these apricots will ensure that the timing is ideal for all parties involved and that the crop will remain at its peak quality.
the apricot problem: case study
Step 4: Inform buyers on East Coast of new production schedule
Figure fastest route for shipping, distribution costs, and scheduling
Since they know when to order months ahead of time, dates for advertising, shipping, handling etc. will be known and accounted for.
the apricot problem: case study
Step 5: Implement MRP system of distribution chain to save buyers money and time
If the fruit is received in a just-in-time system, there will be much less storage time. This will ensure the freshness of the fruit, since it doesn’t have a long shelf life.
