Applied Decision Intelligence & Data Architecture
For Economists · 8 modules · 80 days
For Economists · 80-day technical curriculum

Applied Decision Intelligence & Data Architecture

An 80-day technical curriculum that turns economists into full-stack data architects.

Every day is one page with a video lesson, a timed 2.5-hour plan, the outcomes you should reach, and the full written lesson with step-by-step code. You work in GitHub Codespaces on real Canadian data from the Bank of Canada, Statistics Canada and FRED, and finish with a production pipeline, forecasts and causal estimates.

Start Day 1 Browse the modules
8
modules
80
day lessons
200
hours of guided work
0/80
video lessons published

Video lessons by module

One video per day. Each module has ten.

Published Coming soon

How to use this site

  • Pick a module tab to see what it covers, its tools and its ten days.
  • Open a day to watch the video, follow the timed plan and work through the written lesson. Each day takes about two and a half hours.
  • Videos are being recorded. Until a day's video is published, its page shows a placeholder; the written lesson is already complete.
  • Mark days complete to track your progress. It's saved in this browser.

Part I: Data Engineering Foundations

Days 1–30
Module 1 · Days 1–10

Python for Macroeconomics

Core programming, data wrangling, and API interaction.

  • Days 1–3Syntax, control flow, functions, and virtual environments (.venv).
  • Days 4–7Data manipulation with pandas, handling time-series indexing.
  • Days 8–10Fetching indicators from the Bank of Canada, FRED and Statistics Canada via REST APIs.
PythonpandasrequestspytestGit
0/10 days completed
0/10
Open module
Module 2 · Days 11–20

Analytical SQL with DuckDB

In-process analytical database architecture and query optimization.

  • Days 11–13DuckDB CLI, relational algebra, and loading external CSV/Parquet files.
  • Days 14–17Advanced SQL, window functions, and Common Table Expressions (CTEs).
  • Days 18–20Staging CPI and GDP; year-over-year changes and moving averages directly in SQL.
DuckDBSQLParquet
0/10 days completed
0/10
Open module
Module 3 · Days 21–30

Data Ingestion with dlt

Declarative data pipelines and automated extraction.

  • Days 21–23dlt pipelines, sources vs. resources, and secrets management.
  • Days 24–27Schema evolution, flattening nested JSON, and incremental loading.
  • Days 28–30Resilient extractors for the BoC Valet API and StatCan endpoints, loading into DuckDB.
dltREST APIsDuckDBPydantic
0/10 days completed
0/10
Open module

Part II: Transformation, Validation & Orchestration

Days 31–60
Module 4 · Days 31–40

Data Modeling with dbt

Transforming staged data into production-ready analytical marts.

  • Days 31–33Project initialization, materializations (views/tables), and basic Jinja templating.
  • Days 34–37Modular modeling (staging, intermediate, marts), DRY principles, and dbt macros.
  • Days 38–40A macro data mart joining yield curves with USD/CAD: an open-economy dashboard schema.
dbtJinjaDuckDB
0/10 days completed
0/10
Open module
Module 5 · Days 41–50

Quality Assurance with Great Expectations

Contract-driven data engineering and anomaly detection.

  • Days 41–43Data quality principles, initializing contexts, and connecting to DuckDB sources.
  • Days 44–47Expectation suites, profiling datasets, and automated Data Docs.
  • Days 48–50Validating incoming StatCan data: unemployment rates within 0–100%, continuous dates.
Great ExpectationsDuckDBData Docs
0/10 days completed
0/10
Open module
Module 6 · Days 51–60

Pipeline Orchestration with Prefect

Workflow scheduling, dependency management, and error handling.

  • Days 51–53Flow and task decorators, local orchestration, and state management.
  • Days 54–57Retries, caching, scheduling, and logging.
  • Days 58–60Orchestrating the daily macro-dashboard refresh: dlt → DuckDB → dbt → GX.
PrefectdltdbtGreat Expectations
0/10 days completed
0/10
Open module

Part III: Advanced Analytics & Capstone

Days 61–80
Module 7 · Days 61–70

Forecasting with XGBoost & Chronos

Predictive modeling on macroeconomic time-series data.

  • Days 61–63Supervised learning theory, tree-based models, and univariate forecasting.
  • Days 64–67Hyperparameter tuning, SHAP explainability, and zero-shot forecasting with Chronos.
  • Days 68–70Forecasting Canadian CPI and real GDP growth with features engineered earlier.
XGBoostscikit-learnSHAPChronos
0/10 days completed
0/10
Open module
Module 8 · Days 71–80

Causal Inference with DoubleML

Isolating causal impacts from observational economic data.

  • Days 71–73Causal inference theory, confounding variables, and Directed Acyclic Graphs (DAGs).
  • Days 74–77Nuisance parameters, cross-fitting, and partially linear regression models.
  • Days 78–80The causal impact of BoC overnight-rate hikes on housing starts, controlling for the business cycle.
DoubleMLDAGsstatsmodels
0/10 days completed
0/10
Open module

Final capstone

The course ends with one integrated project: a Prefect-orchestrated pipeline linked to interactive dashboards, built on the architecture of the course's macro dashboard notebook.

ComponentTechnologyCapstone deliverable
Extract & LoadPython, dlt, DuckDBAutomated ingestion of BoC, FRED and StatCan series.
Transform & Testdbt, Great ExpectationsValidated macro marts (yield spreads, real rates).
Model & ForecastDoubleML, XGBoost, ChronosCausal impact models and next-quarter GDP forecasts.
Orchestrate & VisualisePrefect, PlotlyA scheduled, monitored pipeline feeding interactive dashboards.