2. PREDICTIVE ANALYTICS
• The process of extracting trends, behaviors, patterns, and features
from datasets and combinations of data.
• Find relationships in collected data and then predicts the unknown
outcome.
5. The team analyzed retail data and all
possible factors to predict the demand and
set best final price. Non-linear techniques
was used to eliminate noise and extract
relevant features with the best predictive
power (Fig. 1). After that regression analysis
was applied to estimate the optimal
number of purchases for the retailer.
As a result of the analysis of the retail data
the following were able to:
1) Make a demand prognosis based
retail data and various factors such as
season, advertising and marketing
campaigns and economic trends.
2) Establish best final price for the
goods.
3) Develop a predictive model that
allows retailers to make timely goods
orders.
Case Study 1 :
6. Consumer choices are influenced by fashion, lifestyles, past
experiences, income and basic needs. To increase sales, a successful
retailer has to know as much information as possible about every
customer. Structured and analyzed data was perform to customer
segmentation. This example shows a whole customer set divided into
segments based on several parameters (Fig. 2). This information
allows retailers to create an efficient marketing and advertising
campaign based on the information going on in the heads of different
groups of customers.
As a result of predictive modeling for retail:
1) Advanced customer segmentation was performed.
2) Market basket analysis was done to find relationships
between purchases and identify what products different segments of
customers are likely to choose.
3) Found out what products are typically bought together to
benefit the promotional and merchandise strategies of the company
4) Calculated the seasonal demand and frequency of demand
for different products.
5) Created a portrait of an average customer who might be
considering leaving - to improve the company’s loyalty program.
Contd……………..
7. Big retail company was very attentive to their
advertising strategy. Their data were checked to find
out that customer segmentation wasn’t performed
correctly.Analysts held more precise segmentation,
then made micro segmentation and divided all
customers into smaller categories using advanced
clustering techniques such as nonlinear collaborative
filtering and self-organizing maps (Fig.1). Then data
scientists analyzed the data of click conversion and
created a predictive model that anticipated the
patterns of each segment of their customer's
behavior. This helped the company to optimize
online advertising and obtain more customers whilst
spending less on targeted ads.
Case Study 2 :
8. In this case the goal was to analyze campaigns of an advertising
company and see what were the most effective ways to promote
products based on their characteristics and target audience.
Analysts applied a set of multivariate statistical methods (for
example principal component analysis - Fig.2) and offered the
company a predictive model that gave a prognosis on the level of
effectiveness of different types of advertising and PR campaigns.
Main advantages of Predictive Analytics for Advertising:
1) Make advertising campaigns more targeted
2) Perform advanced customer segmentation
3) Increase campaign ROI and profitability
4) Gain actionable insights
5) Support decision making
Contd………….
10. COMPETITIVE ADVANTAGE
• The ability to offer customers products that they specifically need
• Increases sales per customer order.
• Keeping sales channel working with proactive sales calls increases repeat sales and
customer retention.
• Manufacturers can identify sales trends with more accuracy in timing and quantities,
which is particularly valuable when working with long lead times.
• Wholesalers can keep products in-stock while keeping inventory expenses at a
minimum.
• Retailers and restaurants can arrange products and items based on cross-selling and
sales recommendations.
• The ROI for predictive analytics continues to rise as techniques and computing power
continue to accelerate.
11. APPLICATIONS
Business Forecasting Market Clustering
(segmentation)
Product Strategy Customer Analysis Direct Marketing Customer Retention
Cross-Selling Customer Relationship
Management (CRM),
Financial and Economic
Predictions
Risk Management Advertising
Optimization