The role of geography in data integration and predictive analytics 

 Tony Lea, Ph.D. Senior VP and Chief Methodologist, Environics Analytics

In a world where data governance and privacy concerns make it increasingly unable to use data on individuals it is important to have an alternative that works. This paper explains how the use of aggregated data for small scale geography can provide much of the solution in a privacy friendly way. This is called geodemography. After working through some theory, an example is presented that shows how spatial autocorrelation and segmentation can facilitate effective data integration and powerful predictive models. This approach allows users to have thousands of potential predictive variables – socioeconomic and demographic, behavioural, psychographic etc. – beyond those available from customer transactional and tombstone files Not only does this provide better targeting tools but equally important the ability to generate effective marketing communication strategies.

Building reliable machine learning (ML) models for use in production-level systems present specific risk factors not commonly addressed in the practitioners’ literature. As ML continues to play a central role in decision making processes, it is critical to evaluate models under several conditions to identify potential defects. In this session, we introduce model testing and validation methods to ensure the production-readiness of a model. In particular, we discuss concepts such as statistical model criticism, performance uncertainty, model staleness, calibration, algorithmic bias, and interpretability. The underlying methods to tackle these issues are essential for the long-term health of ML production systems, and should be seamlessly integrated in data science pipelines.

See video here https://youtu.be/7YEwoasDVbI

Asmita Usturge, Data Scientist | Microsoft AI

Asmita is a Data Scientist/ Cloud Solution Architect.   She is a passionate data science developer and enabler of cloud technologies for AI.  She works with Microsoft’s high potential customers like financial banks in Toronto to help them transform AI application on Azure cloud platform.    She has 12+ years of experience in the field of computer science, statistics, machine learning and development of AI systems.

The talk:

The AI field is young compared to traditional software development. Best practices and solutions around life cycle management for these AI systems have yet to solidify. This talk will discuss challenges, best practices and technology to operationalize scalable AI models in enterprise

Click here to register for this seminar.

Click here to download the slides from the presentation.

Nataliya Portman | 360Insights.com

Nataliya received her Doctoral Degree in Applied Mathematics from the University of Waterloo in 2010, followed by postdoctoral training at the Neurological Institute in Montreal.

Following her postdoctoral assignment, she developed a novel approach to brain tissue classification in early childhood brain MRIs using modern Computer Vision pattern recognition and perceptual image quality models.

Nataliya has worked in many industries including neuroscience, biotech, the public sector, and various start-up software companies. Throughout her career, she has applied her expertise in Mathematics to develop numerous models including but not limited to machine learning algorithms, computationally efficient algorithms for model validation, and neural networks.

She is the co-inventor of “Bid-Assist”, a strategy for setting up an initial bidding amount to discourage low bidding behaviour, and “AutoVision”, a mobile app that allows automatic taking of pictures of vehicle views and damages recognized by an image classifier.

In January 2019, Nataliya took on a new role in Data Science at 360insights.com, committed to the development of predictive sales channel analytics that will help channel leaders maximize the return on investment of their channel incentive programs.

The Art and Limitations of Accurate Forecasting

Accurate sales forecasting is vital to marketing incentive program design, budget planning and strategic decision making. This seminar will focus on three business use cases:

1. Short and long-term forecasting of target KPI metrics

2. Estimation of sales lift due to promotional program choice and schedule

3. Trade customer/dealer purchasing behavior analysis and prediction of reward winners

Ms. Portman will discuss the challenges of building reliable forecasting time-series models, and how the forecast accuracy can be improved using a multi-model approach. The proposed approach involves preparation of various training datasets containing historic observations of KPIs on different time scales and factors that influence their outcome.

The task of prediction of reward winners based on their to-date and future purchasing activities involves learning of individual buying patterns and estimation of volatility/uncertainty in forecasted KPI values (e.g., volumes of products to be purchased). Knowledge of the uncertainty extent of predicted quantities will reveal individuals with a low degree of uncertainty (candidates to winners). She will show how to effectively evaluate uncertainty per individual that will lead to informed decision making on the final candidate list.