Proposed projects

Project proposed by L’Oréal Singapore

1. Price Optimization in E-commerce stores with Product cannibalization Analysis

Retailers have limited mediums to influence consumer behavior and running price discount has traditionally known for being the easiest to implement. Unlike advertising, price changes can be executed with little to no preparation and yet deliver immediate sales results. However this current practice led to over-discounting and further lower profit margins. It becomes unsustainable for aggressive price discounting without cannibalizing a more profitable product as consumers switch from one brand to another. Students will develop demand models for all products in L’Oréal e-commerce store and come up with an approach to pricing- incorporating their knowledge of product interactions, consumer demand, crawling of competitors’ e-commerce data, study store-wide effects on the level of discount with sales volume and period of running discounts. For further information, please contact QUARK Vivian <Vivian.QUARK@loreal.com>.

2. Data Mining for Store Build Targeting

In order to boost ROI of each store expansion within ASEAN market, L’Oréal needs to develop a deeper understanding of consumer behavior, needs, attitudes and demographics. Students will make use of data mining advances and applications to combine information from various sources such as demographics data (e.g. Age, Education level, Ethnicity, Population density), stores in neighborhood (e.g. distance apart between stores) and internal L’Oréal store data to establish characteristics of each neighborhood. This data-driven segmentations could help identify the potential reach, sales growth of each neighborhood and help to determine new location for store build. Our main focus for this project is one of the ASEAN market (e.g. Indonesia, Vietnam, Thailand, Philippines, India and Singapore) and this depends on availability of census data. For further information, please contact QUARK Vivian <Vivian.QUARK@loreal.com>.

3. Sentiment Analysis on Online Product Reviews Customer reviews are a great source to understanding the likes and dislikes of customers towards L’Oréal products. For e-commerce business, product reviews are very critical as it influences the purchasing decision of new customers in the absence of actual look and feel of the product. Given this, students will utilize machine learning approach for sentiment analysis to classify and analyze human’s sentiments, emotions, and opinions etc. about the products which are expressed in the form of text, star ratings, emoji and the review photos uploaded. Students can further automate an alert to a designated L’Oréal team member of online mentions that requires urgent rectification or a portal for product analytics and brand monitoring. For further information, please contact QUARK Vivian <Vivian.QUARK@loreal.com>.

Project proposed by KLA

1. Build Linearity

Objective: To create an application to improve on Factory Capacity planning

Requirements: The application needs to calculate the system build start date and completion date based on the below given conditions. The production cycle time required to build the systems varies and can be referenced from the MasterOpsPlan. The data from the MasterOpsPlan used is not genuine. It is built up for the sake of testing the logic and functionality of this application.

a) User defines the number of system bays available for Builds.

b) User defines the Build Product to be consider in the calculation.

c) The completion date should generally be 7 days earlier than the Manufacturing commit ship date; However, there may be cases where plan completion dates are less than 7 days or more than 7days. This is inevitable because of the linearization exercise to meet the Manufacturing commit ship date.

d) There should be a toggle button to add an additional requirement to consider system build start date from the 3rd week of every quarter and to be able to disable this when needed.

e) User can input the system completion date when required. When this parameter is defined, the application needs to consider it in the calculation.

f) System bays calculation is based on 1 out and in logic. The out date is reference to the Manufacturing commit ship date. *In special cases if the out date is not reference against the Manufacturing commit ship date then user should have the option to have it reference against the internal ops readiness date in the MasterOpsPlan.

g) User define which system already started and in the pipeline. In this case, the system start date is locked.

h) Application needs to alert user by highlighting builds that are unable to complete on and before the Manufacturing commit ship date.

For further information on any of the above projects, please contact Choo, Felicia (Lee Yan) <Felicia.Choo@kla.com>.

Projects proposed by UBS AG

1. Secure coding gamification

In a world of where very public and very devastating hacks and data breaches are rapidly on the rise, increased software developer awareness of secure coding best practices is crucial to any business' success. Students will create a web game for development teams. Development teams will be able to point the game at a code repository of interest. The game will ingest that code and run best of breed open source security static code analysis tools against the repository. From there, a custom game instance will be created that presents development team with features such as: coding challenges, recorded responses, ratings, top scores, and overall analytics on the security of the repository.

2. Organizational vulnerability risk sssessment

More and more software vulnerabilities are identified each year. When they are announced, organizations prepare rapid responses and expend tremendous resources to close them as quickly as possible and with the least negative impact. However, only 2-5% of the vulnerabilities announced are ever actually exploited. Given this, a risk based approach to remediation could make for a much effective and efficient use of resources. Students will develop an organizational vulnerability risk assessment model and takes into consideration criteria such as: attack vector, attack complexity, exploitation status, vendor, platforms in use by organization, install size, and more.

3. Insider threat visualization

According to a recent Verizon global data breach investigation report, the research team found that approximately 25% of data breaches originated from malicious employees. By monitoring internal activities in an effective way, we can mitigate a quarter of data breach risk. To achieve this objective, the project team will apply cross domain techniques in cybersecurity, data analytics and data visualization to identify potential malicious employees and their activities. The application will visualize employees' activities over a designated period and highlight high risk activities. Furthermore, the application will apply data analytics techniques to identify suspicious activities and prevent data breaches at an early stage.

For further information on any of the above projects, please contact Viramontes, Victor <victor.viramontes@ubs.com>.

Projects proposed by Credit Suisse

1. Optimal trading strategy

The student will learn the optimal trading strategy that provide the minimum expected cost of trading over a fixed period of time. The theoretical framework is minimizing a combination of volatility risk and transaction costs arising from permanent and temporary market impact. The students are expected to develop the model based on the theoretical framework and to test the model performance by using intraday trading data in the stock market. We will guide students on both modeling and testing to complete this project.

2. Catastrophe stress testing

The student will learn to assess the financial risk of a bank under catastrophe events such as earthquake, tsunami, pandemics and flood. The assessment methodology is based on catastrophe risk model framework developed by the insurance industry. The students will develop the model as well as test the model performance by using historical data such as actual losses, direct and indirect impacts on the economy, corporations, and financial industry due to the catastrophes. We will guide students on both modeling and testing to complete this project.

3. Inflection point indicator - Currency crisis

The student will learn to predict the currency crisis of a country. The currency crisis predictive model is developed based on machine learning algorithm on historical financial and macroeconomic data. Currency crisis is defined as currency depreciation of at least 25% over a one month period. Relevant data for various countries are FX rate, external debt - short term, External debt – total, Current account deficit, inflation, foreign-direct investment, portfolio or other investment inflows, foreign currency reserves, level of M2/reserves, real interest rate, GDP, equity Index, export and import. We will guide the students both on the modeling and testing to ensure the success of this project.

For further information on any of the above projects, please contact Chew, Eric <lengsiang.chew@credit-suisse.com>.