UNIT -5
Regression
Model to Forecast Sales
Regression
Analysis forecasting is the most mathematically minded method is usually why
people shy away from it. This technique is meant for those companies that need
in-depth, granular, or quantitative knowledge of what might be impacting sales
and how it can be changed in one direction or the other, as necessary.
Applying
this method successfully requires comprehensive understanding of statistics and
the influences that exert their power on your company’s sales performance.
There are many calculations required to examine relationships between sales and
variables that impact sales.
To
use this you would start to figuring out the reasons you are forecasting,
essentially what it is you want to learn and why that would be valuable. From
there you figure out the factor that is being affected which in this case is
the dependent variable, your sales.
Add
to this the factors that impact the dependent variable, anything that
influences sales. Then, select the period of time you want to review and
collect the data for the variables in question.
Example of Regression Analysis Forecasting
Your business wants to forecast your sales
for the upcoming summer program in order to plan for your budget and figure out
if you need to conduct a second round of hiring for temporary sales reps. In
this scenario, the sales team is the dependent variable and your goal is to
understand what influences it.
So, you compare the sales to an independent
variable, like the number of sales calls. Then you collect data for both the
total seasonal sales and the total seasonal sales calls for the last five
years.
The goal here, again, is to compare what
influences the number of calls had on the number of sales.
Once you set everything up and have the data,
you can get even more granular with that information and review the number of
sales calls as it impacts the number of sales each year, and then again for
each month during the sales season so that you can determine not only how many
new sales reps to hire the following year, but for precisely what months you
need to ramp up seasonal sales reps. Then, you filter them out as the sales
calls and subsequently the sales themselves, start to thin out.
The regression model equation might be as
simple as Y = a + bX in which case the Y is your Sales, the ‘a’ is the
intercept and the ‘b’ is the slope. You would need regression software to run
an effective analysis. You are trying to find the best fit in order to uncover
the relationship between these variables.
 |
| Fig1 |
Pros and Cons
The
upside is that this helps you determine the precise variables that impact sales
at any given time. In other words, this is one of the most accurate forms of
forecasting out there. If you want the real-time data, and only the data, this
is how you get it.
The
downside is that this is accurate, but is incredibly advanced. Be fair warned
that the reason most companies don’t use it or shy away from it, is because not
just anyone can do it. It’s certainly not the easiest method to use.
For
many companies, the variables that have to be taken into account in order to
generate proper forecasts requires someone with a PhD in mathematics to figure
out. This is especially true for larger companies. To that end, the larger
amounts of accurate data is effectively a requirement in order to achieve
meaningful results, and the large amounts can be tricky.
Should You Use Regression Analysis Forecasting?
Regression
Analysis is a highly data driven method which is why it takes skill and regular
practice to do it well. Not only will you need to refine your ability to
execute it, but to understand the results generated therein.
However,
if you are able to properly run your regressions, soon your company will be
able to uncover valuable information about the company that can be used to
drive growth in the future.
Much
like the other methods of sales forecasting, regression analysis may not
necessarily be the optimum solution for your business. To that end, it is
imperative to know how each method works and when it works best in order to
determine if/when it is most suitable for your company.
Moreover,
this does not have to function as a standalone tool; your business might very
well benefit from integrating more than one method particularly if one is a
quantitative method designed to counterbalance and complement a qualitative
method.
Modeling
Trend
Trend
forecasting is a complicated but useful way to look at past sales or market
growth, determine possible trends from that data and use the information to extrapolate
what could happen in the future. Marketing experts typically use trend
forecasting to help determine potential future sales growth. Many areas of a
business can use forecasting, and examining the concept as it relates to sales
can help you gain an understanding of this strategy.
·
Time Series and Trends: Trend forecasting
is quantitative forecasting, meaning its forecasting is based on tangible,
concrete numbers from the past. It uses time series data, which is data where
the numerical value is known over different points in time. Typically, this
numerical data is plotted on a graph, with the horizontal x-axis being used to
plot time, such as the year, and the y-data being used to plot the information
you are trying to predict, such as sales amounts or number of people. There are
several different types of patterns that tend to appear on a time-series graph.
·
Constant Patterns in Data: When looking at
sales numbers, for example, a constant trend is seen when there is no net
increase or decrease in sales over time. The sales may increase or decrease at
specific dates, but the overall average stays the same. However, even if the
average results are the same within a year, there still can be seasonal
changes. For example, sales levels may be consistently greater in the summer
and lower in the winter, although the average is the same in the entire year.
·
Linear Patterns in Data: A linear pattern
is a steady decrease or increase in numbers over time. On a graph, this appears
as a straight line angled diagonally up or down. If someone looked at sales of
VCRs, for example, they might see a diagonal line angled downward, indicating
that sales of VCRs are decreasing steadily over time.
·
Understanding Exponential Patterns: An exponential
pattern is simpler than it may sound. Rather than a slow, steady increase over
time, an exponential pattern indicates that data is rising at an increasing
rate over time. Instead of a straight line pointing diagonally up, this type of
graph shows a curved line where the last point in later years is higher than
the first year, if the rate is increasing. An exponential trend for sales might
indicate that sales were very slow in early years, but the product has grown
increasingly popular each year as more people become interested in purchasing
it.
·
More Complicated Patterns: Trend forecasting
can also deal with patterns that are much more complicated than constant,
linear and exponential graphs. For example, a damped trend may show there was
an overall increase in sales for a number of years and then a sudden stop. A
polynomial trend might show a gradual increase, then stagnation in sales over
time followed by a decrease in sales.
·
Forecasting Using Patterns: Looking at data
over a number of years and finding patterns, you can use this information to
predict future patterns. A trend means the same series of events is happening
over and over. For example, if there is a trend of constant sales each year with
a decrease of sales in winter that is offset by an increase in the summer, a
person might use this pattern to predict that sales will continue to be low in
the winter. Put into action, a store manager might offer additional products in
the winter to help hedge against the expected drop in sales. However,
forecasting isn’t done quickly by just looking at a graph. Forecasters may
translate the a graph’s patterns into a formula to accurately predict what will
happen in the future. They often use spreadsheet software that comes with
built-in trend forecasting tools.
·
Trend Forecasting with Caution: Trend forecasting
is scientific, but it is also uncertain. The longer into the future a forecast
is applied, the more uncertain the results become. Unexpected events can happen
that will disrupt a steady pattern, like stock market downturns changing
consumer behavior and dramatic shifts in users’ access to certain technologies.
The more complicated a pattern appears to be, the more uncertain a trend
forecast is.
Seasonality
Seasonality is a characteristic of a time series in which the
data experiences regular and predictable changes that recur every calendar
year. Any predictable fluctuation or pattern that recurs or repeats over a
one-year period is said to be seasonal.
Seasonal
effects are different from cyclical effects, as seasonal cycles are observed
within one calendar year, while cyclical effects, such as boosted sales due to
low unemployment rates, can span time periods shorter or longer than one
calendar year.
Seasonality Types
There
are three common seasonality types: yearly, monthly and weekly.
(i) Yearly seasonality
Yearly
seasonality encompasses predictable changes in demand month over month and are
consistent on an annual basis. For example, the purchase of swimsuits and
sunscreen prior to the summer months and notebooks and pens leading up to the
new school year.
(ii) Monthly seasonality
Monthly
seasonality covers variations in demand over the course of a month, like the
purchasing of items biweekly when paychecks come in or at the end of the month
when there’s extra money in the budget.
(iii) Weekly seasonality
Weekly
seasonality is a characteristic of more general product consumption and
reflects a host of variables. You may find that consumers buy more (or less) of
different products on different days of the week.
Challenges in estimating seasonality indices
The
seasonality model illustrated here above is a rather naive approach that work
for long smooth seasonal time-series. Yet, there are multiple practical
difficulties when estimating seasonality:
·
Time-series are short. The lifespan of
most consumer goods do not exceed 3 or 4 years. As a result, for a given
product, sales history offers on average very few points in the past to
estimate each seasonal index (that is to say the values of S(t) during the
course of the year, cf. the previous section).
·
·
Time-series are noisy. Random market
fluctuations impact the sales, and make the seasonality more difficult to
isolate.
·
Multiple seasonalities are involved. When looking at
sales at the store level, the seasonality of the product itself is typically
entangled with the seasonality of the store.
·
Other patterns such as trend or product lifecycle also
impact time-series, introducing various sort of bias in the estimation.
Conjoint
Analysis as a Decompositional Preference Model
Conjoint
analysis is one of the most popular tools used for market research purposes. It
is an advanced exploratory technique used to determine how people make
decisions and on what factors do they place real value in various products and
services. It has been widely employed for product/services analysis purposes
since 1970s. Alternatively, this technique is also used for discrete choice
estimation.
The
technique involves analysis of choices people make, and determination of
reasons behind those choices. The tools does not only measure utility of
decision made by consumers, but also helps managers answers questions such as,
“What features should we add to our products, “What impact would changes have
on sales and revenues”, “How will the intended changes give us an edge over
competitors” and “What would be the impact of prices changes on sales”.
Consequently,
the answers to these questions can be used to construct market models, enabling
forecasting in different areas of business. The forecasting model could be
invaluable help to organization as they would help them to plan ahead and
allocate resources accordingly. The conjoint methodology decompositional
approach helps evaluates consumer preference. Consumers are asked to give an
overall rating or score to a particular product profile, which is created by
putting in varying attributes for products and services under scrutiny.
When
customers are about to make a purchase, they are faced with trade-offs in form
of competing products and services. The product exhibited by the competitors
may have varying attributes, which will lead the customer to make a choice
based on those attributes. For example, will the consumers prefer high quality
or low price, Product aesthetics or functionality? In order for businesses to
better understand exactly how customers value various attributes of product and
services, conjoint analysis is a helpful tool, since it provides insights into
the market, which comprises of individual consumers. This technique helps
businesses find the optimal balance between various attributes, the point where
consumers’ value for products and services is the greatest. Conjoint analysis
helps business decision makers determine and quantify the thought process
consumers employ when they are faced with trade-offs between various products
and services available in the market. This is done through quantification of different
aspects or features of products or services. After the analysis is done, and
consumer preferences are worked out, the business is well-equipped to determine
the optimum point where consumer preference and cost to the company is
balanced. At this point business will be able to generate maximum profit, which
is the primary aim of any business.
Conjoint
analysis is different from conventional marketing surveys since it does ask
what products and services consumers prefer, but ask respondents to pinpoint features/attributes
of particular products or services. In most cases, the respondents are asked to
rank the attributes in reference to importance they attract to them. Hence,
this approach offers a more realistic and insightful view of customers’ needs.
In
summation, Conjoint analysis is market surveying technique in which respondents
are required to make choice based on trade-offs between various
attributes/features of product of services. This is usually done by assigning
rating on a scale with upper and lower limits names as ‘Most Preferred’ and
‘Least Preferred.’
Steps in
Conjoint Analysis
Companies
should follow these steps in order to develop a conjoint analysis:
1.
Product/service attributed should be selected. For example, size,
appearance, price, functionality, user friendliness.
2.
Each attribute should be assigned a value. These values of these
attributes will represent the variance available in each of these attributes.
For example, for the attribute of size, the options available could be 5”, 10”
or 25”. Higher value on the scale represent, more importance attached to a
particular attribute.
3.
Product/service should be defined as a combination of all the
varying attributes, which will form the subset of all possible products that a
company can manufacture or deliver.
4.
The combination of attributes through which a product is
represented must be determined. This information can be represented through
visual diagrams, prototypes or mere descriptions.
5.
It must be determined how the results and answers from the
respondents will be aggregated. Decision makers are faced with three choices at
this stage: Individuals responses could be used, the results could be
aggregated to represent a single utility function or the respondents could be
segmentized into sub-groups.
6.
Decision makers at this stage should select the technique they’ll
be employing to process, organize and analyze the data, in order to draw
meaningful conclusion out of the information. A popular model to represent
varying attributes Part-Worth Model. Alternatively, vector models or
ideal-point models could also be used.
Uses of
Conjoint Analysis
Conjoint analyses can break-down large number of attributes into
smaller bundles for evaluations and comparison. There attributes can also be compared
in pairs: Respondents can be asked to indicate preferences between sets of two
or more attributes. In this case, one set of attribute appear on left and
another on the right on the questionnaire. This method is simpler as compared
to evaluating 15 or 25 attributes simultaneously.
A
conjoint analysis is able to breakdown utility to consumer at individual level
as well as aggregate of all the responses. Numerous new techniques have been
recently developed which help companies determine individual level utilities
for choice-based conjoint, which provide companies with useful insight which is
invaluable to the decision making when it comes to marketing, pricing and
product placement.
The
technique offers straightforward methods for experimentation with varying
factors such as price, attributes, price etc. Before a product is launched the
technique helps create a product profile, which can be altered to generate
additional profiles for varying attributes. Consequently this helps businesses
find the balance keeping in view the relative desirability of each alternative
in a choice set and uses each attribute level uniformly throughout the survey.
Conjoint analysis provides information that forms the basis of market
segmentation, whereby a large homogenous market is divided into smaller groups
bases on demographics, preference, age group, etc. Market segmentation enables
businesses to target each homogenous group more effectively since their needs
and preferences are recognized, and decisions are taken according.
Conjoint
analysis can also be employed to exclusive focus on product features and
attributes irrespective of price or brand name, hence enabling calculation of
utility on individual basis in regard to the aforementioned specific features
the companies seeks to evaluate. Moreover, the technique is widely used to
measure the value of brand names in comparison to competing brands. Information
can be obtained as to how strong a particular brand is in comparison to
specific product and price. It helps businesses make decision based on their
brand value in the market, since having a popular brand may not be enough as
changes in price and features could impact demand.
Conjoint
analysis is an important tool which helps in evaluating brand equity and
estimate how market share is impact owing to various tradeoffs between brands,
prices and some specific features. Conjoint analysis can be used to determine
resource allocation, since businesses have already established which attributes
are valued more by the consumers through the analysis. Consequently, companies
can allocate the scare resources accordingly. They can choose to eliminate or
remove features which are superfluous to the customers or not valued by the
consumers. This will save costs for the business, and as already established,
will not impact sales, since these particular products are not valued by the
customers. Reduced cost would lead to more profitability for the businesses.
No comments:
Post a Comment