Module 6

What is STP?
Modern marketing covers various steps of selling goods and
services to customers. There are various techniques and models
that the business analyses and opts for best to sell goods and
services in the market. One very effective marketing strategy is
the STP (Segmentation, Targeting, and Positioning) model.
In STP, S means Segmentation, T means Targeting, and P
means Positioning.

STP marketing example:

Financial Services: A bank (S) segments (STP) by demographics
and behavior, targeting (T) young professionals with a good
income, who are new to investing. They position (P) themselves as
a friendly and educational resource for first-time investors.

What is Segmentation?
Segmentation is the first step of the STP strategy.
Segmentation is the process of dividing the whole market into small
subgroups based on shared characteristics like age, gender, taste,
preferences, etc. Customers having similar needs and behaviours
are to be put together. A market segment is a portion of the whole
market that is expected to respond similarly to a given situation.
Segmentation helps the business identify what type of customers
they should target to sell their product/service. For these reasons, a
company should properly do the segmentation process. Market
segmentation can be done based on:
. Geographic Segmentation
Dividingyouraudiencebasedon region, state, country,etc.
2, DemographicSegmentation
Dividing your audience based on age, education qualification,
occupation, e t e ,
3. Behavioural Segmentation Dividing your audience based
on what they buy, how often they buy , w h a t they browse, ete.

4. Psychographic Segmentation.
Dividing your audience based on lifestyle, hobbies, etc.

Once the company is done with the market segmentation process,

it can focus on choosing the best segment for its products and/or
services. When the segmentation is done correctly, a company can
entirely focus on one or more segments, without wasting any time
and resources.

For example
Volkswagen Group is responsible for producing brands like Audi,
Porsche, Lamborghini, and Ducati. Even after being so expensive,
brands still generate a handsome revenue. Volkswagen depicts the
importance of perfect market segmentation.

What is Targeting?
The process of evaluating market segments and choosing the best
to target comes under Market Targeting. Market Targeting
undertakes the decision of choosing the best target audience and
the degree to which the target market should be targeted. In
simple terms, it is aprocess of choosing the best target
audience for the product/service and declaring the other
segments to be useless for a particular kind of product/service.
A business must determine the target audience after thorough
research; otherwise, the business is going to end up wasting time
and resources with no return on investment. Generally, a new
product/service is first made available to a single target, and if it
remains optimal, the business takes up other segments as well.
Market targeting also depends on the size of the company.
Besides, the more the target markets, the more will the cost
of targeting.

1. Size
Consider how large your segment is.
2 .Profitability
Consider which o f your segments are willing spend t h e m o s t
money o n your product or service
3. Reachability
Consider how easy or difficult it will be for you to reach each
segment with your marketing effort

For example
Nike’s target market includes those people who are interested in
getting fitter.
What is Positioning?
The activity of positioning involves placing the product/service in
the minds of the target customers and making the image of the
product/service superior as compared to other similar products.
Various factors affect the process of positioning such as:
  The larger the size of the target market, the more it will be
difficult to position the product/service.
 If there is no competition in the market, then the business
can create a completely different and new market
positioning strategy.
 If the product has already a good brand value, then it will
be of advantage to the business to position any new
product/service.
 If the company decides to offer fewer prices for its
product/service than the rival firms, then the business can
have an advantage in market positioning.

1. Symbollic Positioning
Enhance the self-image, ago of your customers.
Eg →Luxury car industry, apple i-phone industery, ate.
2. Functional Positoning
Solve your customers problem and provide them with genuine
benefits.
3. Experiential Positioning
Focus on the emotional connection that your customer have with
your products.

For example
Starbucks wanted to make itself ‘The Third Home’ between home
and work so that people can come and relax whenever they are
tired. They decided to target customers with medium and high-
level income. Starbucks uses the following taglines to strengthen
its positioning in the market:
 100% Recycled Paper Use
 The Finest Milk Use
 The Best Coffee
 Rich & Smooth Flavours
 Natural & Clean
Importance of STP Marketing
STP is one of the most important concepts of marketing which
helps in developing a suitable marketing mix. STP is necessary to
understand the behaviour of customers and choose the right target
market for products. It tells the business about the target market
so that they can focus on the right audience and not waste time on
other segments.
Segmentation is necessary to divide the whole customer base
based on their similarities. Segmentation tells us how a particular
segment of customers tends to behave similarly. Targeting is also
a most important activity, as it tells us which customer group the
business should focus on?, and Which customer base is most
appropriate for one’s products/services? Positioning helps the
business in placing a positive image of their products/services in
the minds of target customers so that customers will remember the
product and keep buying it.
Benefits of STP Marketing
 Sharpened Customer Focus: The STP marketing makes
it necessary for you to get into the minds of your
customers. Segmenting your market by their
demographics, needs and behavior will help you clearly
understand who your best clients are. When you know this,
you will be able to concentrate your efforts towards
meeting their needs directly thereby making what you
communicate about resonate with their wants and pains.
 Increasing Marketing Efficiency: There is no more
general approach in marketing. It’s time to channel your
resources to the areas that have potential under STP
marketing strategies. This means that you don’t spend
money where there is no possibility of conversion
henceforth leading into more cost-effective expenditure on
advertisements which will yield better results.
 Boosted Customer Engagement:Picture creating
marketing materials that seem as if they were intended for
individual clients. This is what STP helps you achieve.
Knowing what each of your target groups requires or likes
enables you to make unique messages for them. As a
result, you will have more loyal customers who are deeply
engaged with the brand.
 Distinct Brand Positioning: STP marketing allows for a
different kind of branding in the face of stiff competition.
By deliberately taking a position in relation to the market,
businesses communicate their exclusive selling points to
particular segments. Accordingly, this makes them unique
thereby attracting the right audience while at the same
time keeping off wrong ones.
 Data-Driven Decision Making: To be successful with STP
marketing, you need customer data. You have to divide
your market using this information and then use it to
choose who to sell to and where to sell it. You therefore
make judgments based on facts which minimizes
uncertainty because you only do what your buyers want.
Relationship between Segmentation,
Targeting, and Positioning
Marketing mix undertakes activities like segmentation, targeting,
and positioning, namely STP. These activities are interlinked and
sequenced to form an optimal marketing mix. The very first step is
segmentation. Segmentation involves dividing the whole
customer base according to their needs, preferences, age, gender,
etc. Segmentation tells us how a particular segment of customers
tends to behave similarly.
After dividing the whole customer base into segments, businesses
choose the optimal segment(s) for their products. This process is
known as Targeting. Targeting involves choosing the right
customer segment(s) for the product. After targeting, there
comes Positioning. Positioning involves activities of creating an
image in the minds of the customer base, the base that gets
chosen in the targeting process.
STP Marketing Example
1. McDonald’s
Segmentation – McDonald’s segments its market based on
several factors:
 Demographics: They consider age, gender, income, and
family size. For instance, they have offerings like Happy
Meals for children and Value Meals for budget-conscious
consumers.
 Psychographics: McDonald’s taps into consumers’
lifestyles and personalities. For example, they offer
healthy options for health-conscious individuals and late-
night hours for those seeking convenience.
 Behavioural: They target customers’ buying behaviour,
through the frequency of visits and order preferences. They
have tailored options for regular customers and promotions
to attract occasional visitors.
Targeting – McDonald’s primary target audience includes:
 Families: They provide a family-friendly atmosphere with
play areas and offerings like Happy Meals, targeting
parents and children.
 Young Adults: The menu offers a range of products that
are appealing to young adults, from classic burgers to
trendy items like wraps and salads.
 Teens: They attract teenagers with affordable items, quick
service, and a place to hang out.
 Children: Through Happy Meals, toys, and colourful
packaging, they create an appealing environment for kids.
Positioning: McDonald’s positions itself using the following
strategies:
 Convenience: They emphasize fast service, drive-through
options, and extended hours, positioning themselves as a
quick and convenient dining option.
 Affordability: McDonald’s offers value menus and combo
meals at reasonable prices, targeting budget-conscious
consumers.
 Variety: Their diverse menu caters to different tastes,
from classic burgers to salads and healthier options,
appealing to a broad range of preferences.
 Consistency: McDonald’s maintains a consistent quality
and taste across its global locations, creating a sense of
familiarity and reliability.

K means clustering notes (sir)

Discriminant analysis
It is a statistical technique used in machine learning and
multivariate statistics for classifying a set of observations into
predefined classes. It models the difference between two or more
classes and assigns new observations to one of those classes based
on their features. It is primarily used when the dependent variable is
categorical and the independent variables are continuous (although
categorical variables can sometimes be used).

Here are key types and details of Discriminant Analysis:

1. Types of Discriminant Analysis

• Linear Discriminant Analysis (LDA):

• Assumes that the classes have a common covariance matrix
and that data is normally distributed within each class.
• It creates linear boundaries (or decision surfaces) between the
classes.
• Often used when there are two or more classes and is
particularly useful for dimensionality reduction.
• Example Use Cases: Face recognition, medical diagnosis,
customer classification.
 • Quadratic Discriminant Analysis (QDA):
• Similar to LDA but does not assume equal covariance matrices
for the classes, leading to more flexible and non-linear decision
boundaries.
• Useful when the assumption of equal covariance among
classes doesn’t hold, though it can be more prone to overfitting
with small datasets.
• Canonical Discriminant Analysis:
• Focuses on describing the differences between multiple groups
and finding the axes that maximize the separation among
predefined groups.

Difference between

Discriminant Analysis and Cluster Analysis are both techniques

used in data analysis, but they differ significantly in their purpose,
approach, and use cases. Here’s a detailed comparison of the two:

1. Purpose

• Discriminant Analysis:
• Goal: It is primarily used for classification and prediction.
• Supervised Learning: Discriminant Analysis works with
labeled data, meaning that the classes (or groups) are known
beforehand.
• Example: Classifying whether a customer will buy a product
based on their demographic and behavior data. Here, you
already know the categories (“buyer” or “non-buyer”).
• Cluster Analysis:
• Goal: It is used for grouping data points into clusters based on
their similarities.
• Unsupervised Learning: In Cluster Analysis, there are no
predefined labels. The algorithm discovers groupings within the
data based solely on the inherent structure of the data.
 • Example: Grouping customers into segments based on their
purchase history, where the segments are not known
beforehand.

2. Approach

• Discriminant Analysis:
• Modeling the Difference: It focuses on finding a boundary or
discriminant function that separates predefined classes based on
a set of input features.
• Assumptions: Linear Discriminant Analysis (LDA) assumes
that data from different classes is normally distributed with
equal covariance matrices (for LDA).
• Example Use Case: Predicting whether a tumor is benign or
malignant based on a set of medical features.
• Cluster Analysis:
• Finding Groupings: It identifies clusters (groups) of similar
objects or data points without any prior knowledge of group
membership. It groups data points to minimize intra-cluster
distances (similarity) and maximize inter-cluster differences.
• No Assumption of Labels: Unlike Discriminant Analysis,
Cluster Analysis works without labels, making it exploratory.
• Example Use Case: Segmenting customers into groups based
on buying behavior without predefined categories.

3. Nature of the Problem

• Discriminant Analysis:
• It is supervised and deals with classification problems. You
need to know the categories in advance and use them to train
the model.
• Cluster Analysis:
• It is unsupervised and deals with grouping or clustering
problems. The goal is to find natural groupings in the data.

4. Methods Used
 • Discriminant Analysis:
• Examples include Linear Discriminant Analysis (LDA) and
Quadratic Discriminant Analysis (QDA).
• It produces decision boundaries for classification and evaluates
how well the features separate known classes.
• Cluster Analysis:
• Techniques include K-means clustering, Hierarchical
clustering, DBSCAN, etc.
• Focuses on minimizing intra-cluster variance (similarity within
a group) and maximizing inter-cluster variance (differences
between groups).

5. Example Scenarios

• Discriminant Analysis:
• You have a labeled dataset of customer data, and you want to
predict whether a new customer will churn based on their profile.
• A bank wants to classify loan applicants as low, medium, or
high risk based on known historical data.
• Cluster Analysis:
• You want to explore a dataset of customer purchases to
identify natural segments (e.g., high spenders, budget-conscious
shoppers) without predefined categories.
• Analyzing genes based on expression profiles to find similar
groups for research purposes.

Summary Table
Aspect Discriminant Analysis Cluster Analysis
Purpose Classification and prediction Grouping and segmentation
Supervision Supervised (requires labeled data) Unsupervised (no labeled data needed)
Goal Separate known classes Discover natural groupings
Key Techniques LDA, QDA K-means, Hierarchical, DBSCAN
Example Use Cases Predicting customer churn, disease classification Customer
segmentation, pattern recognition

To summarize: Discriminant Analysis is used when you have

predefined classes and want to classify new data points, while
Cluster Analysis is used to explore and identify patterns or groups
in unlabeled data.
Difference between
Discriminant Analysis and Logistic Regression are both used for
classification problems in statistics and machine learning, but they
differ in terms of their assumptions, modeling approaches, and
appropriate use cases. Here’s a detailed comparison:

1. Purpose and Approach

• Discriminant Analysis:
• Goal: It is used for classifying observations into predefined
categories based on a set of predictor variables.
• Linear Discriminant Analysis (LDA) assumes that the
predictor variables for each class are normally distributed with a
common covariance matrix.
• Quadratic Discriminant Analysis (QDA), a variant, allows
for different covariance matrices for each class, resulting in
more flexible boundaries.
• Approach: It models the distribution of the predictors
separately within each class and uses Bayes’ theorem to
estimate the probability that a given observation belongs to
each class.
• Logistic Regression:
• Goal: It predicts the probability of belonging to a particular
class (e.g., binary outcome like 0 or 1) based on predictor
variables.
• Approach: It directly models the probability of the outcome as
a function of the predictor variables using a logistic function
(sigmoid curve). Logistic regression estimates the coefficients of
a linear combination of predictor variables to maximize the
likelihood of observing the given data.

2. Assumptions

• Discriminant Analysis:
• Assumes that the predictors follow a multivariate normal
distribution for each class (especially for LDA).
 • LDA assumes equal covariance matrices among classes,
whereas QDA allows for different covariance matrices.
• Logistic Regression:
• Makes no distributional assumption about the predictors.
• Assumes a linear relationship between the log-odds of the
response variable and the predictor variables.
• It does not require equal variance-covariance matrices across
groups.

3. Data Requirements

• Discriminant Analysis:
• Works best when the number of observations is much larger
than the number of predictor variables (to ensure robust
covariance matrix estimation).
• Typically performs better when the normality assumption holds
and the data is well-separated.
• Logistic Regression:
• Can handle a wider variety of data distributions since it does
not assume normality.
• More robust when the predictor variables do not meet strict
distributional assumptions.
• Works with smaller datasets more effectively because it does
not need to estimate covariance matrices.

4. Interpretability

• Discriminant Analysis:
• Provides linear or quadratic decision boundaries based on how
the predictor variables separate the different classes.
• LDA can be used for dimensionality reduction by transforming
predictors into fewer dimensions that maximize class
separability.
• Logistic Regression:
 • Easy to interpret the coefficients since they represent the
change in the log-odds of the outcome for a one-unit change in a
predictor, holding other predictors constant.
• More interpretable in terms of the impact of predictor variables
on the probability of an outcome.

5. Use Cases and Suitability

• Discriminant Analysis:
• Suitable when you know or can reasonably assume that the
predictors are normally distributed within each group.
• Can be more effective than logistic regression when the groups
are well-separated and normally distributed.
• Often used in fields like medical diagnosis, finance, and pattern
recognition when the normality assumption holds.
• Logistic Regression:
• More commonly used, especially for binary or multinomial
classification problems, as it does not require stringent
assumptions about the predictor variables’ distribution.
• Effective for problems with categorical predictors, binary
outcomes, or when interpretability of the predictor effects is
important (e.g., marketing analytics, healthcare modeling).

6. Handling Nonlinearity

• Discriminant Analysis:
• Linear Discriminant Analysis creates linear decision
boundaries. Quadratic Discriminant Analysis allows for non-
linear decision boundaries by relaxing the assumption of equal
covariance matrices.
• Logistic Regression:
• Non-linearity can be introduced by adding polynomial terms
or using transformations of the predictor variables.
• Can incorporate interaction terms easily.

Summary Table
Aspect Discriminant Analysis Logistic Regression
Model Type Generative Discriminative
Assumptions Normally distributed predictors per class No distributional assumptions
Covariance Matrix Assumption Equal (LDA), Different (QDA) Not required
Data Type Works well for continuous predictors Works well with continuous and categorical
predictors
Complexity Linear or Quadratic boundaries (LDA/QDA) Flexible, linear model with possible
extensions
Use Cases When normality and equal covariance assumptions hold, well-separated groups
Broadly applicable for binary/multinomial outcomes

Key Differences in Practice

• Discriminant Analysis focuses on modeling the distribution

of predictors within each class, then uses these models to classify
observations. This makes it effective when assumptions hold but
less flexible when they don’t.
• Logistic Regression directly models the probability of the
outcome and is easier to use in many practical settings due to
fewer assumptions and greater flexibility. It remains one of the
most popular tools for binary classification problems.

In summary, Discriminant Analysis is ideal when you have

normally distributed data and want to classify into well-defined
categories, while Logistic Regression is more broadly applicable
and easier to interpret for classification problems with fewer
assumptions.

Hierarchial clustering
Dendogram
Two group da
Wilks lambda def