Strategy

Capital Intensive ⋅ Industrials

Prescriptive ⋅ Mixed Integer Optimization

Impact: High ⋅ Feasibility: Low

This module presents how corporate strategy teams use analytics in the boardroom. The module uses prescriptive analytics to support a disguised multinational conglomerate re-position its portfolio of business units. The case is applicable across problem statements that seek to maximize net present value of a portfolio including investments, oil and gas assets, or business units.

Modeling COVID-19 Progression and Policy Effects

Public Sector ⋅ Healthcare

Prescriptive ⋅ Epidemiological Model

Impact: High ⋅ Feasibility: High

In the Spring and early summer of 2020, all eyes were on policy makers to choose how and when to re-open broader society. While sophisticated sounding terms like epidemiological model and exponential growth entered the common lexicon, it was clear that there were no easy answers. Embrace uncertainty and follow along the process to build a COVID-19 case prediction model.

Real Estate Pricing

Capital Intensive ⋅ Real Estate

Predictive ⋅ Optimal Regression Trees

Impact: Medium ⋅ Feasibility: High

Real estate makes up a significant aspect of the economy, and it can also makes up a significant amount of our individual net worth. While we expect real estate pricing to be tied to square feet or square meters, there are variables that will not manifest linear effects like waterfront property and the number of bedrooms or bathrooms. In this module, we reconcile this issue.

Movie Recommendations

Technology ⋅ On-Demand Streaming

Descriptive ⋅ Clustering

Impact: Low ⋅ Feasibility: High

Competition between subscription video on demand companies is fierce. These services differentiate themselves by offering customers movie recommendations based on their preferences and viewing history. This module demonstrates how to use clustering to make movie recommendations for subscription video on demand services' customers.

Sentiment Mining with Twitter

Commerce ⋅ Telecommunications

Predictive ⋅ Natural Language Processing

Impact: Low ⋅ Feasibility: High

So many Tweets, so much emotion. In today’s world, business entities must stay current with their customers, competition, and the ubiquitous "influencers". But who has time to keep up with all that volume? Learn how corporate entities use natural language processing to track user sentiment of the "Twitterverse".

Blue Bikes

Commerce ⋅ Transportation

Descriptive ⋅ Clustering

Impact: Low ⋅ Feasibility: High

Bike sharing programs use clustering to segment their riders and identify and engage personas. The complication is that standard clustering methods are complex and confusing. Leverage the state-of-the-art method, interpretable clustering to identify personas in a logical and straight forward manner.

Default Risk

Financial Services ⋅ Banking

Predictive ⋅ Optimal Classification Trees

Impact: Medium ⋅ Feasibility: High

Leveraging data and analytics to predict the likelihood of default of a loan is not a new concept. Being able to justify a decision to approve a loan to both regulators and consumers is a new concept. Engage this module to master two emerging methods: intrinsically interpretable models and bolt-on explainable tools.

Interpretable Predictive Maintenance for Turbofans

Capital Intensive ⋅ Energy

Predictive ⋅ Optimal Survival Trees

Impact: Medium ⋅ Feasibility: Low

As complexity of industrial systems grow, understanding the ways they fail becomes more challenging. Leverage analytics to find the right moment to perform maintenance so components are not prematurely replaced.

Hard Drive Failure

Commerce ⋅ Computer Hardware

Predictive ⋅ Optimal Decision Trees

Impact: Medium ⋅ Feasibility: Low

The Internet of Things (IoT) is the future. The first IoT sensors were found in hard drives. In order to demonstrate the true potential of IoT data, we provide an IoT anchor case that demonstrates how to harness the data to make investment and maintenance decisions by analyzing data center IoT sensors.

Market Model

Capital Intensive ⋅ Mining

Predictive ⋅ Model Driven Prediction

Impact: Medium ⋅ Feasibility: Medium

Predicting future product pricing and market demand is a challenging task that is sometimes best suited for a model driven analytics product. Whereas a data driven product starts with data, then fits a model, and finally tests for objective accuracy, a model driven approach starts with a series of assumptions of the world, the environment or human behavior, and then adds data to create a projection. While market models are built by experts, business managers have a critical role in supporting the effort.

Framingham Heart Study

Public Sector ⋅ Healthcare

Predictive ⋅ Logistic Regression

Impact: High ⋅ Feasibility: Low

Doctors seem to really care about their patient’s cholesterol and blood pressure levels. This has not always been the case. Build a logistic regression model to predict coronary heart disease and learn how the Framingham Heart Study changed the world.

Interventional Radiology Scheduling

Public Sector ⋅ Healthcare

Prescriptive ⋅ Integer Optimization

Impact: Low ⋅ Feasibility: High

Computers make scheduling fair (and more profitable). Whether setting shifts at a restaurant, the class schedule in a school, or transportation routes at an airline, decision makers must walk a persistent line between the efficiency and individual desires. Join an interventional radiology team as it leverages mixed integer optimization to achieve both higher revenue and a fairer scheduling environment.

Immigration Enforcement

Public Sector ⋅ Legal

Predictive ⋅ Difference Trees

Impact: Medium ⋅ Feasibility: High

Immigration policy is a contentious issue in the United States. In this module we provide data-driven facts to support the policy debate by clearly answering whether the Immigration and Customs Enforcement agency (ICE) is meeting guidance provided by the Department of Homeland Security. We leverage a natural experiment to help any higher headquarters entity understand if its’ subordinate elements are meeting their mandate.

Handwriting Recognition

Technology ⋅ Computer Vision

Prescriptive ⋅ Deep Learning & Optimization

Impact: Medium ⋅ Feasibility: Low

Computers can read. But the key question is "how well?" Join a research and development effort to train a computer to read historic documents. During the ride, also pick up key skills on how to manage a R&D effort with key output metrics in order to ensure the project stays on the rails.

Air Force Tanker Routing

Capital Intensive ⋅ Aerospace & Defense

Prescriptive ⋅ Integer Optimization

Impact: Medium ⋅ Feasibility: Medium

Geopolitics are complex. Hot spots ebb and flow. Access to airspace is granted then restricted. Even more complex is maintaining a supply chain of fuel assets to ensure that fighter aircraft can serve their mission in these dynamic environments. While sophisticated optimization engines exist to support mission planning, without expert guidance, their output will be meaningless. Learn to be the expert guide.

Automated Market Research at Tinker Air Force Base

Capital Intensive ⋅ Aerospace & Defense

Predictive ⋅ Semi-Supervised Learning

Impact: Low ⋅ Feasibility: High

Aerospace maintainers often rely on a set of small of vendors to execute niche services on their highly complex aircraft engines, components, and airframes. If these small vendors close their shops, the maintainers must scramble to find alternative sources. Fortunately, these maintainers can use advanced analytics to do this work for them so they can focus on turning wrenches.

Capital Expenditure

Capital Intensive ⋅ Mining

Prescriptive ⋅ Mixed Integer Optimization

Impact: High ⋅ Feasibility: Low

Complete the critical steps of an operational transformation by building a capital expenditure (CAPEX) model to augment the previously built universal supply chain model.

Supply Chain Optimization

Capital Intensive ⋅ Mining

Prescriptive ⋅ Mixed Integer Optimization

Impact: High ⋅ Feasibility: Low

The initial steps of a operational transformation include reducing costs to improve profit margin, in order to earn the right to grow. This disciplined approach of first saving money, and then smartly re-investing that money is logical. It is also challenging. A core element to meet this challenge is unity of effort across a given enterprise. An underlying universal supply chain model can not only drive value, but also keep the organization aligned.

Capacity Planning

Capital Intensive ⋅ Construction

Prescriptive ⋅ Robust Optimization

Impact: Medium ⋅ Feasibility: Low

In business, we have become accustomed to providing ranges instead of point estimates when we predict project costs, production quantities, financial returns, and consumer demand. Ranges communicate our certainty or lack thereof and point estimates are nearly almost incorrect. The field of robust optimization enables us to input ranges into our modeling efforts instead of point estimates. Join this module to conduct factory capacity planning with robust optimization.

Talent Management

Entertainment ⋅ Baseball

Predictive ⋅ Linear Regression

Impact: Medium ⋅ Feasibility: Medium

You probably know the Moneyball story. But, do you know the math? In this module, apply linear regression and build an algorithm to select player personnel in a manner that will outperform the scouting system just like the Oakland As did in the early 2000s.

Structuring Data to Predict Stroke

Public Sector ⋅ Healthcare

Descriptive ⋅ Natural Language Processing

Impact: Medium ⋅ Feasibility: Low

It is estimated that 80% of healthcare data is in an unstructured format. For example, everyday neurologists around the world review CT scans and MRIs and dictate their findings which are recorded in reports. Computers cannot analyze this type of unstructured data. Join this module to explore cutting edge natural language processing methods capable of translating radiology reports to actionable intelligence to improve patient outcomes.

COVID-19 Infection and Mortality Calculator

Public Sector ⋅ Healthcare

Predictive ⋅ XGBoost

Impact: Low ⋅ Feasibility: Medium

In April 2020, the medical community was seeking novel solutions to a host of COVID-19 related problems. One example: COVID-19 tests were time consuming, expensive and require patients to visit facilities in person. Another example: clinicians could not accurately predict which patients would need critical care. Follow an MIT research team build rapid solutions in sparse data conditions.

Jury Selection

Public Sector ⋅ Legal

Predictive ⋅ Bias Detection

Impact: High ⋅ Feasibility: High

In 1986, the U.S. Supreme Court ruled that using race as a reason to remove potential jurors is unconstitutional. Despite this ruling, a large disparity in juror strike rates across races appears to remain.

Insuring Capital Equipment

Capital Intensive ⋅ Aerospace & Defense

Prescriptive ⋅ Simulation

Impact: Medium ⋅ Feasibility: Medium

Are you comfortable with your insurance policies? Are you properly covered? Are you over paying for scenarios that will never manifest themselves? Or, do you have no idea because the policies are too confusing? If so, you’re not alone. Join an airline as it uses Monte Carlo Simulation to set its fleet insurance policy.

Congenital Heart Surgery

Public Sector ⋅ Healthcare

Predictive ⋅ Optimal Classification Trees

Impact: High ⋅ Feasibility: Low

Engage the next milestone in surgical outcome analysis. Release the assumptions that pre-operative risk factors are linear and additive and follow along a new technique to measure the effectiveness of individual hospitals at performing a series of congenital heart surgeries.

D2Hawkeye

Public Sector ⋅ Healthcare

Predictive ⋅ Logistic Regression

Impact: High ⋅ Feasibility: Low

In healthcare, they say an ounce of prevention is worth a pound of cure. What if the healthcare systems could identify patients before a major health complication and intervene? Join the D2Hawkeye story and learn how a clever use of a penalty matrix in predictive modeling can dramatically improve the ability to identify high risk patients and save lives.

Opioid Prescription

Public Sector ⋅ Healthcare

Predictive ⋅ Anomaly Detection

Impact: High ⋅ Feasibility: Low

Prescription opioid abuse is a severe public health crisis in modern United States history. This module demonstrates how to use machine learning to identify illegal diversion of controlled substances by licensed, consumer-facing buyers, such as pharmacies and doctors. With external data, we discover complex patterns in the transactions of controlled substances across the prescription drug supply chain in the United States.

Surgical Risk

Public Sector ⋅ Healthcare

Predictive ⋅ Optimal Classification Trees

Impact: High ⋅ Feasibility: Medium

Emergency surgery, which deals with acute threats to life, accounts for roughly 7% of surgeries in the United States. It is critical for surgeons to be able to reliably estimate the risk of mortality and complications prior to committing to surgery. Evaluate a clinical decision support tool.