UNIT-3 DATA SCIENCE

1)LOGISTIC REGRESSION
What is Logistic Regression?
• Logistic Regression is a technique used in business analytics to predict outcomes.
o For example: Yes/No, Success/Failure, Churn/Stay, Approve/Reject.
• It looks at different factors (like customer behavior, product usage, demographics) and
calculates the probability of an outcome.
• Unlike other methods that predict numbers, logistic regression predicts likelihoods.
Types of Logistic Regression
1. Binary Logistic Regression
o Predicts two possible outcomes (Yes/No).
o Example: Will a customer churn or stay?
2. Multinomial Logistic Regression
o Predicts more than two outcomes, but outcomes are not ordered.
o Example: Predicting the preferred product category: Electronics, Clothing, or
Groceries.
3. Ordinal Logistic Regression
o Predicts more than two outcomes with a natural order.
o Example: Customer satisfaction levels: Low, Medium, High.
Importance of Logistic Regression in Business Analytics
• Helps make decisions based on probability rather than guessing.
• Can identify key factors that influence outcomes (like what causes customers to
leave).
• Supports marketing strategies by targeting high-risk customers or potential buyers.
• Useful in risk management, like predicting loan defaults or insurance claims.
• Improves resource allocation by focusing on customers who need attention.
Detailed Example: Predicting Customer Response to a Marketing Campaign
Scenario:
A company is launching an email marketing campaign for a new product and wants to know
which customers are likely to respond (buy the product or show interest). This helps target
only high-potential customers, saving costs and improving ROI.
Step 1: Define the outcome
• Outcome is binary: Respond (Yes) or Not Respond (No).
Step 2: Identify influencing factors
Factors that might affect response:
• Customer’s past purchase history
• Age group
• Location or region
• Engagement with previous emails (opened/clicked)
• Type of subscription (basic, premium)
Step 3: Collect and analyze data
• The company collects data from past campaigns: who responded and who didn’t.
• Observed patterns might include:
o Customers who opened previous emails are more likely to respond.
o Premium subscribers respond more than basic subscribers.
o Certain age groups (e.g., 25–35) show higher response rates.
Step 4: Predict customer response
• Logistic regression is used to calculate the probability of each customer responding.
• Example: A customer has:
o Past purchase = yes
o Age = 28
o Premium subscription
o Opened previous emails
The model predicts a 70% probability of responding.
Step 5: Take action
• The marketing team focuses only on high-probability customers, e.g., probability >
60%.
• Benefits include:
o Reduced marketing costs
o Higher conversion rates
o Better ROI on campaigns
………………………………………………………………………………………………….
2) DECISION TREE
What is a Decision Tree?
• A Decision Tree is a tool used in business analytics to make decisions based on data.
• It breaks down a complex decision into smaller, easy-to-understand steps in a tree-like
structure.
• Each “branch” represents a decision rule, each “node” represents a condition, and the
“leaf” shows the outcome.
• It can be used for classification (predicting categories, like Yes/No) or regression
(predicting numbers).
Importance of Decision Trees in Business Analytics
• Easy to interpret: Visual and simple to understand.
• Helps in decision-making: Shows which factors influence decisions the most.
• Handles both numerical and categorical data.
• Useful in sales, marketing, risk management, HR, and finance.
• Can detect important patterns in customer behavior.
…………………………………………………………………………………………..
3) RANDOM FOREST
What is Random Forest?
Random Forest is a type of ensemble learning technique used in data analytics and machine
learning.
• It builds multiple decision trees and combines their predictions to get a more accurate
and stable result.
• Each tree in the forest is trained on a random subset of data and a random subset of
features, which reduces overfitting and improves prediction accuracy.
Importance of Random Forest in Business Analytics
• High accuracy: Aggregating results from many trees reduces errors.
• Handles large datasets: Works well even with big data and many variables.
• Feature importance: Shows which factors influence outcomes the most.
• Versatile: Can be used for classification (yes/no, categories) and regression (numeric
predictions).
• Robustness: Less affected by outliers or missing data.
Random Forest Applications
• Customer churn prediction: Businesses can use random forests to predict which
customers are likely to churn (cancel their service) so that they can take steps to retain
them. For example, a telecom company might use a random forest model to identify
customers who are using their phone less frequently or who have a history of late
payments.
• Fraud detection: Random forests can identify fraudulent transactions in real-time.
For instance, a bank might employ a random forest model to spot transactions made
from unusual locations or involving unusually large amounts of money.
• Stock price prediction: It can predict future stock prices. However, it is important to
note that stock price prediction is a very difficult task, and no model is ever going to
be perfectly accurate.
• Medical diagnosis: These can help doctors diagnose diseases. For example, a doctor
might use a random forest model to help them diagnose a patient with cancer.
• Image recognition: It can recognize objects in images. For example, a self-driving car
might use a random forest model to identify pedestrians and other vehicles on the
road.
…………………………………………………………………………………………………