Transcript Prediction Method

Application of the RFID Data
Mining to an Apparel Field
Professor Kesheng Wang
Department of Production and Quality Engineering
Norwegian University of Science and Technology
N-7491 Trondheim, Norway
Tel. 47 73 59 7119, Fax 47 73 59 7117
E-mail: [email protected]
RFID data ming, KDL, NTNU
1
10-Apr-16
Outline

Introduction
 Prediction method
 Data format
 Experiment
 Conclusions
 A new project proposal
RFID data ming, KDL, NTNU
2
10-Apr-16
Introduction
• This project proposes a new method that efficiently uses
the RFID data collected from apparel shops.
• This method learns prediction models from the data by
using data mining techniques. The models represent
relationships between the number of sales in the next
term and the actions of customers, such as the number of
pick-up, the number of fitting, the number of customers,
and so on.
• It is possible to predict sales volume by applying the
present RFID data to the models.
• This project verifies the efficiency of the method through
numerical experiments based on the RFID data collected
from two branches of an apparel company.
RFID data ming, KDL, NTNU
3
10-Apr-16
Prediction Method
• The prediction method is based on the RFID data.
• The data is composed of independent variables and
a dependent variable. The independent variables
correspond to the number of customers, stock, sales,
and so on in a week.
• The method applies the models to the RFID data in
this week and predicts the number of items sold in
the next week. The managers can decide the number
of items ordered in this week by referring to the
number of stock in this week and the predicted
number of sold items.
RFID data ming, KDL, NTNU
4
10-Apr-16
Prediction Method
NTNU
RFID data ming, KDL, NTNU
5
10-Apr-16
A format of training example
Indipendent Variable
No of
pick-up
Time of
pick-up
Depend
ent
Variabl
e
No of
fitting
Time of
fitting
RFID data ming, KDL, NTNU
No of
custome
rs
No of
stock
(whole)
6
No of
stock
(shop)
No of
stock
(backya
rd)
10-Apr-16
No of
sales in
previou
s term
No of
saløes
in
present
term
No of
sales in
the next
term
Conclusions
•
•
•
•
•
•
•
This project proposed a method that predicts the number of sales
in the next term based on the RFID data.
The experimental results show the possibility of the prediction,
even if it is necessary for the prediction models to be revised their
performance.
In future work, we will tackle on the improvement of the
prediction models.
We will try to collect new RFID data. These improvements of the
data can revise the models.
On the other hand, we will aggressively tackle to establish
many methods which the RFID data efficiently activates in
various fields.
RFID data ming, KDL, NTNU
7
10-Apr-16
A new peoject proposal:
Title: Assessing changes in quality of
perishable produce in chilled supply
chains using RFID logged data
Partners:
1. NTNU (Data ming and system integration)
2. Manchester University (RFID Temperature sensors,
Logistics)
3. Hrafn (RFID, SCM)
4. ……
RFID data ming, KDL, NTNU
8
10-Apr-16
Motivation and Drivers
NTNU
Temperature-related loss in quality of
perishable produce is significant. Not only
does this quality represent a cost, but also it
generates waste produce that is expensive to
dispose of. As a result, there is an economic
incentive to estimate the temperature-related
loss, which would subsequently enable a
quality-control strategy based on temperature
monitoring.
RFID data ming, KDL, NTNU
9
10-Apr-16
Objectives
NTNU
To identify and/or develop a technique that
can predict changes in quality of in-transit
perishable produce using logged
temperature.
RFID data ming, KDL, NTNU
10
10-Apr-16
Expected End Result

A data mining technique that uses logged
temperature to estimate the loss in quality
of perishable produce in transit.
 Develop a BIP project with NFR
RFID data ming, KDL, NTNU
11
10-Apr-16
Framework/plateforem
RFID data ming, KDL, NTNU
12
10-Apr-16
Case study: predicting Remaining Shelf Life
(RSL) for chilled sea food

RFID data ming, KDL, NTNU
13
10-Apr-16
Experiment results
NTNU
RFID data ming, KDL, NTNU
14
10-Apr-16
RFID Setup in Lab.
NTNU
Components
Company
Price
Temperature sensor array
RFID Tag
RFID Reader
RFID Antenna
Middelware (software)?
Others?
RFID data ming, KDL, NTNU
15
10-Apr-16