Parcel Delivery - Intel Segmentation and Last Mile Opt.pdf

Copyright © SAS Institute Inc. All rights reserved.
Lojistikte SAS

Müşteri
Kullanıcı
FORTUNE Global 500®
aralıksız Büyüme
Gelir
Gelirin Ar & Ge
SAS Hakkında

Computer
Vision Forecasting and
Optimization
Machine Learning,
Deep Learning
Natural Language
Processing
Data
Management
Visualization
Decision
Management
Deployment
AI
Türkçe Mevcut
ORACLE, SAP
Entegrasyonu Mevcut
Robotik Proses Otomasyonu ve
Karar Motorları ile Entegre
SAS ML ve AI Yetkinlikleri
AutoML dahil
AutoForecasting ve hiyerarşik
tahminleme dahil

COVID Döneminde
Bir Örnek Uygulama
Intelligent Segmentation and
Last Mile Optimization

Innovation: Intelligent Segmentation and last mile optimization
Demand shift has disrupted the shopping cart and hence need to revisit the planning and
optimization of last mile delivery
Priority
Deliver Values when
it is needed most Last Mile Optimization
with COVID19
Utilize data ,get more
accurate customer
demand profiling, delivery
optimization
Discover new retail
distribution challenges
under COVID19
Intelligent Optimizations
driven planning and delivery

Retail Segmentation and route optimization
Senior citizen orders grocery Waits for 2 weeks
Order is delivered in 15 days
1. Complication
2. How SAS can help
• The COVID 19 has highlighted for the world population an important need and an important obligation to be respected,
namely the isolation of people to avoid widespread infection of the virus.
• In this scenario, red areas are defined, that is, particular areas in which there are strong movement restrictions. The
primary needs of the population remain unchanged and therefore it is of primary importance to be able to meet the
demand for products by area.
• Consider the above scenario where a senior citizen orders grocery online in a COVID-19 crisis hit region since he cannot go
out due to COVID-19 outbreak.
• Grocery which was expected to be delivered in 1 day, took 15 days to be delivered. Due to this long wait some senior
citizens are facing a lot of problems in pandemic hit areas.
• SAS can help business and the consumer both by proposing analytics-based recommendation which can reduce the
delivery time by using advanced machine learning models and can help the community as a whole

Retail Segmentation and Route Optimization
Segment the product order into various segments using priority rules and machine learning algorithm-
based segmentation.
Optimization algorithm can be used to optimize the product delivery and answer some of the possible
solutions to deliver product:
• What products should be delivered to customer address.
• What all products can be delivered to a common pit stop where someone from customer’s family
can picked up the order.
• What is the most optimized route to be followed for a group of orders.
Order would be bifurcated into red areas where there is a strong movement restriction and non red
areas order in terms of priority.
Products would be delivered into low and high priority products. For example, order received from
senior citizen would have high priority. Products which require food with particular food
intolerance/allergies like lactose, gluten etc would be given a higher priority.
Applying machine learning algorithm-based segmentation we can segment the products into different
product bundles, each having a specific priority over others.
SAS Optimize engine would be used to further optimize the product bundle to be delivered and the
route to be followed. The route would be optimized such that maximum population can be covered
with the products.
Challenge
Approach
Results
Data
• Location Data
• Customer Data
• Order Data
• Store Data
• Product Data
• Inventory Data
• Delivery availability data
What are the products to be delivered in red areas?
What are the product bundles as per different priority to be delivered?
What is the best optimized route to be followed to deliver these products?

No Data type Data description
1 Location Data Information of street, city, province, postal code, etc. of the customer
2 Customer Data Customer name, age, family size, other person if available to pick up the grocery, food allergies, etc.
3 Order Data Name, description, quantity, etc.
4
Stores hierarchy and their
attributes
Id, name, description, address, coordinates, opening date, closing (or reconstruction) date, working
hours, format, square, regulatory restrictions (e.g. no alcohol in store, no tobacco in store), etc.
5
Products hierarchy and their
attributes
Id, name, description, merchandise hierarchy, attributes
6
Planned assortment +
assortment history
Planned assortment for the period of analysis (including test sales dates)
Level: SKU-store-day
7 Planogram history POGs data: store, SKU, number of facings, length of facings, etc
Different Types Of Data Useful For Analysis

Modeling Approach and Benefits
• Catering the need of people affected in
COVID-19 especially senior citizens.
• Optimizing the priority of order that needs
to be delivered.
• Optimizing and prioritizing between
perishable and nonperishable goods.
• Optimizing the delivery for retail store so
that they can cater maximum population in
optimized cost and time.
• Using analytics-based decision making with
increased efficiency.
• Clean the data and
prepare modeling
dataset
Data

What We Can Do Together
• Clean the data and
prepare modeling
dataset
Data
We will help on data prep integrated
to core systems, databases and ERP

We will provide demand forecasts and
predictions that will be input to optimization

Interactive analysis of Optimization results for multiple
scenarios will be shared on Desktop and mobile

We will have all results, optimized routes, alternate
scenarios, last mile delivery status and ETA’s in dashboards
for management and for vendors, as well as clients

Bir Diğer Örnek Uygulama
Price Optimization For Trucking

Company Confidential – For Internal Use Only
Overview
Price Optimization
Segmentation
Establish customer willingness to pay
at the line item level. Microsegments
derived from attributes which explain
the variability of price.
Tasks:
• Define price metrics
• Identify & bin attributes
• Select significant attributes
• Build segmentation hierarchy
Demand Modeling
Estimate the relationship between rates
and demand response to rate
changes. Understand customers expected
reaction to price at the segment level.
Tasks:
• Time series forecast of shipments
• Price elasticity statistical modeling
Optimization
Formal optimization to select a set of
decisions that optimize the objective
while adhering to business rules.
Tasks:
• Create optimization price grid
• Identify objective function and
business constraints
• Build mathematical optimization
model
15

SAS Price Optimization for Trucking
Different User Roles
Revenue Operator
• Protect & Improve
Profitability
• Develop Segmentation
Models
• Create Peer Group
Analysis with Floor &
Ceiling Price Levels
• Manage GRI & MPC
Pricing Manager
• Oversee Account
Activity &
Performance
• Create and Monitor
Performance Alerts
• View Peer Group
Analysis and Select
Target Price
Account Manager
• Sales volume oriented
• Monitor Performance
Alerts
• Customer Service
• Marketing, campaigns

Demonstration
• Existing Customer Quote (example)
ü View most frequent Customer Behavior by Core Service
ü Select customer Segmentation Criteria
ü View Floor-Target-Ceiling prices within Peer Group

Select Core Service – based on
Transaction Count by Core Service
frequency
18

Observe most frequent Customer Behavior (right
pane) and select Segment Criteria
19

Example: Monitor & Price Adjustment
Compare Actual to
Commitment
Performance based
on multiple factors
We will have real time discount
recommendations based lot size
drop size, origin - destination

Demonstration
• Set Price Change Events - Optimize
ü Run and Compare Different Scenarios
• Analyze and Evaluate the Impact - View Rate Card by
ü Shipping Zone
ü Weight
ü Scenario

Price Optimization
Run and compare different scenarios
22

Price Optimization
View Rate Card by Shipping Zone and/or
Weight and/or Scenario
23

A full-service freight transportation provider in the
Northeast. Company has grown from a small local
carrier into the largest privately held less-than-
truckload (LTL) company in the United States.
LTL Truck Price Optimization
Business Needs
Update a manual pricing process via Excel
Deliver analytically derived price recommendations
to Account Managers responsible for thousands of
rate quotes
Provide a custom user interface for ease of use,
adoption, and increased job performance
How SAS® supports the project:
Development services for a stand-alone solution.
Functional design, business consulting, process
flow, and user interface design services.
Segmentation and price validation with customer
through UAT.
Included ‘alert’ reporting of low performing customer
transactions and a ‘customer value’ report.
We’ll fix this for you!

SAS Referansları

https://www.youtube.com/watch?v=FATv5nOtazM

https://www.sas.com/en_us/insights/articles/analytics/ups-loves-logistics.html

https://www.sas.com/en_za/customers/relais-colis-eng.html

3 Hızlı Kazanım Önerisi

Kargo
Analitik
Akademisi
Veri analitiği yazılım ve
çözümlerine ilişkin eğitimler
ve PoC çalışmaları
SAS Master Data
Management ve
Data Quality
Çözümleri
Veri kalitesi çözümleri ve
Veri kataloglama
yaklaşımlarımızı sunacağız
SAS Unified
Insights
Teklifi
Veri analitiği platfromu
çözümleri
Şirketinize en uygun paket

sas.com
Teşekkürler
Altan.atabarut@sas.com

Parcel Delivery - Intel Segmentation and Last Mile Opt.pdf

More Related Content

Parcel Delivery - Intel Segmentation and Last Mile Opt.pdf