Official Profile & Portfolios
The W App (2023—)
W is the flagship product of World Ranking, Inc. (“WR”), a crowd-source ranking institute and private corporation.
W Quickstart Guide:
Note:
If you’re looking for a long-term investment in a startup, then look no further. Our resilient capital structure, commitment to high product standards, and relatively low operating costs; give us every assurance that WR is built to last. The yearly net earnings are estimated at $77 million (Level I projected after-tax net earnings, for 1 million MAUs, optimal scenario, post-stabilization; the details on Level II and III revenue chains are unavailable at this time, but we expect them to surpass our Level I, which forms the base revenue).
You may request a copy of our investor information package by contacting us directly at info@ranktheglobe.com, with your legal name, contact information, business details (if applicable), and any other info of relevance. Class A, B and C shares are available for purchase to eligible investors.
David’s Portfolio of Conceptual Art & Design
Selected original artworks of David span acrylic paintings (e.g. the mythic golden eagle), sketches (e.g. the maglev car), and digital models (e.g. the future city)
Note: Below is an archived page of David’s Academic/ Early Professional portfolio, which is no longer in use.
Disclaimer: By downloading, accessing or using any content from this website, you agree that you are doing so at your sole risk, and for educational purposes only
Section Navigation:
Selected Corporate Projects & Exhibits
- Southlake-Ontario Health Cancer Screening Test of Change (to be deployed provincially), GWR Auto-Prediction Results Published via Interactive App, Region: SCOHT Area (pipeline programmed in Python), 2022-2023
Selected from a number of comparable test-of-change and AI projects across the province and hailed as “The Gold Standard” by the Ontario Government, the state-of-the-art spatial prediction models and associated automated components were designed and built by David, the sole expert who represented Southlake & Ontario Health (SCOHT) at multiple interagency committees and working groups. Among the participants were ESRI and Carleton University Department of Health Informatics. The models enabled the automatic GWR modeling of any designated urban region, which resulted in local predictor significance (adaptively classified by the proxy statistic β coefficient/SE). β/SE is a GWR proxy for local predictor significance; its statistical thresholds are used to determine the level of significance (-2 to 2), which is made conceptually intuitive to the users (e.g. “Statistically Relevant (-)”). Positively correlated (+): e.g. + income, + screening rate; Negatively correlated (-): e.g. + income, - screening rate. Local Significance adapts to the unique characteristics of the neighborhoods (or Dissemination Areas) through the GWR spatial and local weightings (e.g. adaptive bandwidth/Gaussian), and does not necessarily imply that a DA should adapt to the overall significance in a specific direction of a relationship (i.e. Simpson’s Paradox). “Locally significant” means that the predictor is showing a high significance (i.e. systematic non-random pattern) on the dependent variable (e.g. Rate of Cancer Screening) at the given Dissemination Area in the given direction of a relationship (+ or -). In short, local interventions in such areas are deemed relevant with a high magnitude of effect relative to any predictors not deemed relevant/not selected. Study designs such as RCT trials, which account for temporal precedence and theoretical plausibility, could be then used to determine causal relationships, guiding actionable local interventions.
- GWR Auto-Prediction Results Published in Interactive Web App - Region: East Toronto Unincorporated Area, 2022-2023
The models were tested, independently evaluated, and successfully replicated in another urban region, which produced adaptive results and new insights for policymakers; particularly worth noting was that urban areas were better generalized by the existing ONMarg Indicators than areas of suburban/rural setting. GWR performance also scales with the unit of measurement (generally, the smaller the better, e.g. block groups, if available, tend to produce more substantial results than dissemination areas)
- Penn Exhibition Website: Urban Real Estate Valuation Tool (Standalone & Portable) (programmed in Python), 2021
This adaptive spatial prediction model David designed allows RE investors to identify high-return RE investment opportunities in adjoining neighborhoods pending population growth in any designated urban region, in addition to generating reliable assessment figures for insurance companies and tax agencies (parameters such as regional inflation rate, 10-year risk-free interest rate, and construction cost psf can be auto-detected from the model inputs and plugged into predefined equations, given correct field names).
at the MUSA program of the Weitzman School (2020-2021)
Python, JS, R and SQL are the primary languages used
Selected Advanced RE Finance & Investment coursework (real-world based)
at the Wharton School (2020-2021)
†Sharing of models is currently prohibited, please contact Wharton Real Estate Department for permission to view/evaluate
at UWaterloo, Specializations: City Planning, GIS & Real Estate Development (2014-2019)
-
Supervised Machine Learning: Surface Object Detection: Segmentation & Classification (SVM) (programmed in Python) Mar 2021
-
Supervised Machine Learning: Everyday Object Classification (K-NN & SVM) (programmed in Python) Feb 2021
-
Detecting NDVI and urban structures on Woody Island using satellite bands (programmed in JS) Feb 2021
-
Global Real Estate Development Suitability Model (programmed in JavaScript) Dec 2020
-
Heuristic/Unsupervised Machine Learning: Clustering Analysis of Chicago, IL, Population Characteristics (programmed in Python) Dec 2020
-
Heuristic/Unsupervised Machine Learning: K-Means Clustering, Philadelphia, PA (programmed in R) Dec 2020
-
Point Pattern Analyses (NN, K-Functions), Philadelphia, PA (programmed in R) Dec 2020
-
SLR Real Estate Damage Prediction Model for New York City, NY (programmed in Python) Dec 2020
-
Supervised Machine Learning (Random Forest): Single-Family Home Price Prediction for Philadelphia, PA (programmed in Python) Dec 2020
-
Machine Learning: Single-Family Home Price Prediction for Miami, FL (programmed in R) Dec 2020
-
NDVI and Urban Network Analysis, Philadelphia, PA (programmed in Python) Oct 2020 – Nov 2020
-
Spatial Lag, Spatial Error, GWR Median House Price Prediction for Philadelphia, PA (ArcGIS + R) Nov 2020
-
Global Cumulative Travel Cost Estimation (programmed in JavaScript) Oct 2020
-
Supervised Machine Learning: Home Repair Tax Credits Benefit/Cost Prediction (programmed in R) Oct 2020
-
Supervised Machine Learning: Narcotics-related Crime Prediction, Chicago, IL (programmed in R) Oct 2020
DAVID’S PORTFOLIO OF SPATIAL INFORMATICS, GIS & RE FINANCIAL MODELING (© 2014-2021)