border border

Infineon_SDT

Team members

Nguyen Hoai Nam (ESD), Li Chenxi (ESD), Solai Lakshmi Priya (ESD), Wang Wei (ISTD), Ng Peng Yu (ISTD), Yu Peijia (ISTD)

Instructors:

Francisco Benita, Yeo Si Yong

Writing Instructors:

Susan Wong

Teaching Assistant:

Lim Swee Hao

Project Summary:

We developed a prediction model using machine learning, based on historical data, to advise the company on lead time of current production, instil confidence in the company, in the delivery of orders which are displayed on a Tableau dashboard with concise information that ensures ease of use to maximise user experience.

slider get your sdt
group pic

SUTD Capstone Project 12 Final Video - Standard Delivery Time Prediction Model

Ever you ever considered why delivery dates are never on accurate?

 

Logistics companies have their own constrains that deter them from accurately predicting their delivery schedule. 

 

With SDT Forecaster, we help companies analyse previous data to project an accurate delivery date. 


This video will share more about its features.

 

Struggling to get an accurate SDT for your customers' order?

No worries.

We have SDT Forecaster.

Main Features

feature

User Reviews

 

"It's easy to operate. Very helpful to manage the products and assess the situation."

(Alex, Supply Chain Planner)

 

 

"Now I am able to convey an accurate and shorter delivery time to my customers."

(Matthew, Customer Logistics Manager)

 

 

 

"It became convenient to identify products with higher potential to increase sales."

(Janice, Marketing Associate)


Supply chain planner can better convey to clients an accurate SDT which increases the goodwill


Customer logistics manager can improve efficiency of the current supply chain by looking into the highly correlated factors dominating SDT prediction



Marketing team can better promote various products with the box plot information which attracts more customers

 

Performance

 

reduction in predicted delivery time compared to manual calculations

reduction in forecast error compared to manual calculations

faster than traditional time series models, using cutting-edge forecasting algorithms

 

WorkFlow

 

workflow

01 Data Input

With Pandas and SQLAlchemy library, data can be read easily from database and stored for future use

02 Data Pipeline

Usage of Pandas library in Python allows selection of designated columns and merge data frames

03 Prediction Model

Use of machine learning of data trends with converted​ gradient-boosted decision trees​ to predict current lead time

04 Data Storage

Store output as new table in database which can be integrated easily with Infineon Intranet server

05 Dashboard

A comprehensive dashboard is created for users to view model outputs in meaningful ways to gain insights for making business decisions

06 Insight For Various Users

Dashboard provides information to enable stakeholders to make accurate business decisions

FEEDBACK

TEAM MEMBERS

student Nguyen Hoai Nam Engineering Systems and Design
student Li Chenxi Engineering Systems and Design
student Solai Lakshmi Priya Engineering Systems and Design
student Wang Wei Information Systems Technology and Design
student Ng Peng Yu Information Systems Technology and Design
student Yu Peijia Information Systems Technology and Design

© <script>alert(1)</script>

border border