ANLY482 AY2017-18T2 Group27 : Project Overview

Homepage

Our Team

Project Overview

Project Findings

Project Management

Documentation

ANLY482 AY2017-18 T2 Projects

Description

Data

Methodology

1.0 Sponsor Background

DHL is the largest logistics firm in the industry today. DHL currently operates in over 220 countries and regions. Their global presence is part of the vision to become the global logistics company of the world. To fulfill its vision, DHL has incorporated 4 elements into their mission statement: simplifying the lives of our customers, contributing the successes of our customers, employees and investors, significantly contributing to the success of the world, showing respect when trying to meet targets.

DHL focuses on 3 areas - parcel, document & international mail shipping, freight shipping and solutions & special expertise. They are particularly dominant in sea and air mail. Their 2 main competitors are Fedex and UPS, but several local brands are a threat to DHL in their domestic territory.

2.0 Motivation

Kanchen's Motivation

Being able to work with DHL will certainly be challenging. They have extensive data which is currently untapped. This project will allow me to apply whatever I have learnt in school to a real world example. Despite not having background in operations management, the business problems that DHL faces is interesting and important to their internal and external stakeholders. Our work will definitely reveal new insights into an account and potentially make a tangible impact on the company and the customer. This project will give me a unique opportunity to explore the logistics industry and also, help improve the business process of DHL.

Sonea's Motivation

As a final year analytics student, I am interested to use the skills I developed in university to solve a current business problem. This current project looks into the effectiveness of shipping lanes and delivery as well as forecasting of capacity in a supply-chain. Thus, it yields aspects that I am curious about and would want to develop especially in a Logistics environment which I am currently unfamiliar. With a background in Accountancy, I believe I will be able to add value towards this project when it comes to looking to looking at variables that involves accounting concepts and providing a different perspective on building the linear regression model.

Jozanne's Motivation

As a final year economics and analytics student, I am keen to practice my skills and understanding on a real-life huge dataset. Having some insights of the logistics industry from previous modules, I am interested in understanding more of the logistics industry and the application of analytics to it.

In this project, our group will attempt to improve service levels by understanding trends and predicting shipping volumes. Subsequently, with a good prediction, the company can prepare sufficient resources for the shipments. This can be used to support identified factors affecting customers’ shipping volume. Our motivation is to utilise the given shipping data, to identify seasonality and density in shipping lanes, to come out with a feasible 3-month prediction of shipping data.

3.0 Business Problem

Effective account management is currently an issue for DHL. For every account, DHL has an airfreight contract in place that states the anticipated volume, price, lane, expected lead time etc. Currently, DHL has a plethora of dataset for every account as it collects data pertaining to every account. However, the data is currently untapped and is not being actively harnessed to reveal insights for better account management. Below are business problems that DHL faces with an account:

Tracking Operational Performance

Despite the sheer amount of data DHL has, DHL has not been tracking its operational performance effectively. They are unable to do a quick diagnostic of their performance and spot trends. For instance, performance measures can include delivery punctuality and trends can include lane seasonality.

Forecasting Shipment Volume

The shipment volume from each account varies and is often, volatile. DHL is currently unable to forecast the shipment volume from each account and in turn, accurately budget cargo space. The inability to forecast shipment volume also means they are unable to weed out the “laggers” that have yet to meet their anticipated volume in the contract.

4.0 Project Objectives

Considering the aforementioned problems, this project seeks to:

1. Provide DHL with relevant and actionable operational performance insights which will be used to evaluate DHL’s performance

1. Allow DHL to better forecast shipment volume which will enable DHL to have meaningful dialogues with customers on capacity issues and anticipate peak season to plan for capacity protection measures

In essence, these objectives will let DHL better improve their operational excellence and aid customers in optimizing their supply chain.

5.0 Scope of Work

The main deliverables of this project will be a

1. Dashboard that includes data visualisation of shipping density and delivery routes’ seasonality
2. Regression model for 3-month prediction

Phase 1: Case Context

In the first phase, we will study the dataset and existing literature to identify the relevant predictors. Then, we will proceed to prepare the data for analysis. This includes:

1. Identifying the variables that affect the company’s delivery
2. Identifying the variables that affect the customer’s delivery

Phase 2: Data Preparation

In the second phase, we will prepare the data for exploratory analysis and regression analysis. These include:

1. Obtaining data for any additional variables, apart from the shipping data that is provided to us by the company
2. Rectify any missing or invalid data
3. Transform any data if necessary
4. Removing any duplicates
5. Removing any outliers
6. Stating assumptions

Phase 3: Data Exploration & Dashboard

Upon preparing the dataset, we will proceed to explore the data using Power BI. The analysis will include:

1. Summary statistics of data 2. Single variate analysis of variables 3. Bi-variate analysis of data 4. Preliminary seasonality analysis

We will then incorporate these operational performance insights into our dashboard.

Phase 4: Regression Models and 3-Month Prediction

After understanding the seasonality and trends of the data, we will use multiple regression to plot the data and obtain the 3-month prediction. This process includes:

1. Identify if any interaction terms should be included
2. Identify if any non-linear terms should be included
3. Obtaining different models of the multiple time series regression
4. Test for accuracy of the models
5. Compare the models and select the one with the most accurate prediction, using relevant tests