Statistics and actuarial sciences
The purpose of this project is to develop three predictive models for Temenos’ Customer Intelligence offering including: (1) Customer Attrition; (2) Loyalty Scoring; and, (3) Next Best Offer. Customer attrition, also called “churn”, or “defection”, is a major business issue for organizations to address. It is crucial to maintain and grow customer relationships in order to sustain profitable growth.
Multivariate random effects model for Integrated measurement of green veneer thickness and roughness
Thickness and surface roughness are the two main veneer quality criteria affecting material recovery, plywood glue bond quality, and glue consumption. At present, on-line green veneer thickness and surface roughness are generally not measured, which causes difficulty in assessing overall veneer quality for a better control of log conditioning and veneer peeling process.
A quantitative risk prediction model is to be constructed. We need to determine if the available data will fit an existing model and validate the results or if a new statistical model is required. Each case will be allocated into one of three categories (low, moderate and high risk). This stratification must have clinical validity andutility. The cut-offs for the stratification will be established based primarily on clinical utility and on the availability of the data. The cut-offs will be optimized to achieve optimum AUC, NPV, PPV, sensitivity and specificity values.
Fleetmetrica is a new business startup that offers an innovative and patent-pending technology called SafetyMonitor for monitoring and controlling fleet safety of large commercial vehicles. Fleetmetrica has achieved success with initial fleets, including improved driver habits using their product and are interested in formally quantifying the effectiveness of their technology.
For banks and financial institutions data is one of the most important assets, providing the key to true differentiation and competitive advantage. Insight Business Intelligence, solutions offered by Banking Software Company Temenos, turn figures into meaningful information for business decision making. However the existing business rules based models could be improved by using novel statistical techniques to capture the stochastic dependence in the data.
Investigate parametric machine learning algorithms to develop anomaly detection methods on real-time data
The industry partner, Metafor is developing a new class of IT system management solution. As part of this project, Metafor is building a product feature that monitors computer and network activities and looks for signs of anomalies. This is an important problem as anomalies are usually associated with abnormal user or system behaviors that can potentially result in problems such as system breakdown.
Health economic modeling of Ulipristal Acetate in the treatment of uterine fibroids in the Canadian setting
Given the public health-care system in Canada, many drugs are paid through by the provincial drug plans. Our government makes decisions on which drug to pay for through a process known as health technology assessment whereby manufacturers submit a document that details the clinical, economic and other aspects (e.g. ethical, social) relating to their drug.
Assessment of the uncertainty in calculating the wind farm energy production using statistical methodology
The evaluation of the wind resource and its associated energy production over the life cycle of the wind power project is one of the most important elements to determine the profitability of a wind power project since a bank bases its decision on the energy production that is estimated to occur with a 99% probability. Its evaluation is a highly uncertain process and it requires the assessment of the uncertainty in the measurement of the wind resource.
The project will assess the financial risks to DGAG (Desjardins Groupe d’assurances générales) associated with payments of accident insurance claims. A large database is available on the losses incurred due to different aspects of insurance claims (medical costs, rehabilitation and attendant care, etc.), and this project will assist DGAG in developing exposure assessments for future accident benefit claims. By adopting a Bayesian statistical approach, the uncertainties associated with various data sources and modelling assumptions can be integrated into a single, coherent framework.
As a part of the Million Tree Challenge, the MITACS intern will work with ReForest London staff and partners to conduct research to determine how many trees are planted in the City of London this year. As a part of this work, they will conduct of review of the best data collection practices of other cities that are or have attempted to plant a large number of trees. They will determine what data to collect and how it will be collected, and develop, test, and refine a statistical model for using this data to estimate the total number of trees getting planted each year.