Evidence field guide / DS-36

Data Science Learning Path

A hands-on path from consequential questions to reproducible decision products.

01Reader: 5-7 minutesUnderstand the question, visual argument, caveat, and decision.
02Practitioner: 20-40 minutesComplete the guided analytical task and inspect its receipt.
03Researcher: 1-3 hoursReproduce, diagnose, and extend the method.

Path calibration36 modules / 6 waves

method

W01Evidence foundations
W02Reason under uncertainty
W03Credible comparisons
W04Time, space, and systems
W05Modern analytical systems
W06Publish a decision product

decision readiness

Wave 01 is in viewEvidence foundationsFrom a broad claim to clean, versioned evidence.

Canonical progression / M01-M36

Follow the evidence, not the software menu

The complete route is visible from the start. M01 is open now; every planned module keeps its outcome, prerequisites, source IDs, and deliverable in view without offering a dead control.

Categorical key

Eight analytical tracks

Foundations
Statistical reasoning
Communication
Econometrics
Dynamics
Structures
Modern analytics
Publishing

Wave 01
Evidence foundations
From a broad claim to clean, versioned evidence.
6 modules
M01Available now
Foundations / Reader
Question → measurable claim
Outcome
Turn ‘Germany is losing competitiveness’ into a testable metric tree, reproduce its peer-median productivity estimate, and bound the conclusion.
Prerequisites
Starting point
Datasets
DS001, DS003, DS004, DS011
Deliverable
Article + notebook
M02Planned
Foundations / Reader
Provenance, license, and versions
Outcome
Trace a popular chart to its original series and record vintage, query, license, and checksum.
Prerequisites
M01
Datasets
DS001, DS002, DS063, DS076
Deliverable
Article + manifest
M03Planned
Foundations / Practitioner
Tidy data, keys, and joins
Outcome
Build a reliable country-year analytical table from multiple sources.
Prerequisites
M01, M02
Datasets
DS001, DS002, DS012
Deliverable
Code-along
M04Planned
Foundations / Practitioner
APIs, JSON, and SDMX
Outcome
Build a reusable, cached fetcher with clear parameters and failure states.
Prerequisites
M02, M03
Datasets
DS003, DS004, DS006, DS007
Deliverable
Code-along
M05Planned
Foundations / Reader
Units, deflators, PPP, and indexes
Outcome
Choose and justify the transformation that matches a claim.
Prerequisites
M01, M03
Datasets
DS001, DS002, DS008, DS013
Deliverable
Article + notebook
M06Planned
Foundations / Practitioner
Missingness, flags, and outliers
Outcome
Diagnose missingness and choose transparent treatment rather than silent imputation.
Prerequisites
M02, M03
Datasets
DS004, DS014, DS021, DS059
Deliverable
Lab note
Wave 02
Reason under uncertainty
Describe distributions, sampling, uncertainty, and visual integrity.
7 modules
M07Planned
Statistical reasoning / Reader
Descriptive statistics that answer questions
Outcome
Match the summary statistic to the decision.
Prerequisites
M05, M06
Datasets
DS001, DS010, DS014
Deliverable
Article + exercise
M08Planned
Statistical reasoning / Practitioner
Distributions, inequality, and concentration
Outcome
Build Lorenz, Gini, top-share, and concentration measures with caveats.
Prerequisites
M07
Datasets
DS007, DS016, DS068, DS076
Deliverable
Code-along
M09Planned
Statistical reasoning / Researcher
Sampling and survey weights
Outcome
Use final and replicate weights and communicate sampling uncertainty.
Prerequisites
M06, M07, M08
Datasets
DS047, DS048, DS052, DS053
Deliverable
Research notebook
M10Planned
Statistical reasoning / Practitioner
Uncertainty and bootstrap intervals
Outcome
Create intervals and distinguish their sources.
Prerequisites
M07, M08, M09
Datasets
DS002, DS059, DS065
Deliverable
Article + notebook
M11Planned
Statistical reasoning / Practitioner
Tests, effect sizes, and power
Outcome
Report effect size, uncertainty, and minimum detectable effect.
Prerequisites
M09, M10
Datasets
DS015, DS047, DS052
Deliverable
Code-along
M12Planned
Statistical reasoning / Researcher
Bayesian updating and Monte Carlo simulation
Outcome
Combine a prior, evidence, and simulated outcomes without overstating certainty.
Prerequisites
M10, M11
Datasets
DS032, DS059, DS080
Deliverable
Research note
M13Planned
Communication / Reader
Chart grammar, integrity, and accessibility
Outcome
Choose scale, encoding, annotation, colors, and alternatives ethically.
Prerequisites
M05, M06, M07, M08, M09, M10
Datasets
DS024, DS030, DS042
Deliverable
Visual clinic
Wave 03
Credible comparisons
Move from conditional relationships to explicit causal designs.
9 modules
M14Planned
Econometrics / Practitioner
Regression as controlled comparison
Outcome
Specify, diagnose, and interpret a transparent baseline model.
Prerequisites
M07, M10
Datasets
DS001, DS011, DS024
Deliverable
Article + notebook
M15Planned
Econometrics / Practitioner
Functional form, logs, and interactions
Outcome
Interpret elasticities, nonlinearities, and interaction effects correctly.
Prerequisites
M14
Datasets
DS012, DS021, DS064
Deliverable
Code-along
M16Planned
Econometrics / Researcher
Panel data and fixed effects
Outcome
Estimate fixed effects, cluster uncertainty, and state the identifying variation.
Prerequisites
M14, M15
Datasets
DS004, DS011, DS024
Deliverable
Research notebook
M17Planned
Econometrics / Reader
Causal diagrams and confounding
Outcome
Express a causal question as a DAG and identify adjustment traps.
Prerequisites
M14
Datasets
DS001, DS015, DS023
Deliverable
Article + simulator
M18Planned
Econometrics / Practitioner
Experiments and A/B tests
Outcome
Design power, allocation, guardrails, and analysis for a realistic intervention.
Prerequisites
M10, M11, M17
Datasets
DS015, synthetic extension
Deliverable
Lab note
M19Planned
Econometrics / Researcher
Difference-in-differences and event studies
Outcome
Test parallel trends and interpret dynamic effects.
Prerequisites
M16, M17
Datasets
DS004, DS024, DS025
Deliverable
Research notebook
M20Planned
Econometrics / Researcher
Regression discontinuity
Outcome
Assess bandwidth, manipulation, continuity, and local interpretation.
Prerequisites
M14, M17
Datasets
DS005, administrative extension
Deliverable
Research note
M21Planned
Econometrics / Researcher
Instrumental variables
Outcome
Explain relevance, exclusion, and weak-instrument risk.
Prerequisites
M14, M17
Datasets
DS023, DS029, teaching simulation
Deliverable
Research note
M22Planned
Econometrics / Researcher
Synthetic control and counterfactuals
Outcome
Build, validate, and placebo-test a synthetic comparison.
Prerequisites
M16, M17
Datasets
DS001, DS002, DS012
Deliverable
Capstone method
Wave 04
Time, space, and systems
Model dynamics, geography, remote signals, and connected structures.
6 modules
M23Planned
Dynamics / Practitioner
Time series: trend, cycle, and autocorrelation
Outcome
Transform, diagnose, and decompose a time series.
Prerequisites
M05, M10
Datasets
DS006, DS008, DS025
Deliverable
Article + notebook
M24Planned
Dynamics / Practitioner
Seasonality and structural breaks
Outcome
Detect and explain recurring cycles and discontinuities.
Prerequisites
M23
Datasets
DS025, DS030, DS069
Deliverable
Lab note
M25Planned
Dynamics / Practitioner
Forecasting and rolling validation
Outcome
Backtest forecasts, quantify uncertainty, and separate scenarios from predictions.
Prerequisites
M23, M24
Datasets
DS002, DS025, DS045
Deliverable
Article + notebook
M26Planned
Structures / Practitioner
Geospatial analysis and MAUP
Outcome
Join spatial data, choose a projection, and test geographic sensitivity.
Prerequisites
M03, M06, M13
Datasets
DS036, DS037, DS038
Deliverable
Geo code-along
M27Planned
Structures / Researcher
Remote sensing as economic evidence
Outcome
Process raster data and validate a proxy against ground data.
Prerequisites
M26
Datasets
DS029, DS034, DS039
Deliverable
Research notebook
M28Planned
Structures / Researcher
Networks, flows, and input-output dependencies
Outcome
Build a network, calculate centrality/concentration, and distinguish gross from value-added flows.
Prerequisites
M03, M08
Datasets
DS016, DS018, DS061
Deliverable
Article + tool
Wave 05
Modern analytical systems
Evaluate prediction, clustering, language, retrieval, and monitoring.
5 modules
M29Planned
Modern analytics / Practitioner
Classification, calibration, and asymmetric costs
Outcome
Evaluate discrimination, calibration, imbalance, and decision thresholds.
Prerequisites
M10, M14
Datasets
DS032, DS055, DS072
Deliverable
Code-along
M30Planned
Modern analytics / Practitioner
Clustering and principal components
Outcome
Standardize, reduce dimensions, cluster, and test stability.
Prerequisites
M08, M13
Datasets
DS052, DS053, DS063
Deliverable
Lab note
M31Planned
Modern analytics / Practitioner
Text as data
Outcome
Tokenize, compare frames, and validate a text measure.
Prerequisites
M06, M10
Datasets
DS054, DS062, DS075
Deliverable
Article + notebook
M32Planned
Modern analytics / Researcher
Embeddings, retrieval, and semantic maps
Outcome
Build and evaluate retrieval rather than displaying a pretty embedding cloud.
Prerequisites
M30, M31
Datasets
DS063, DS067, DS076
Deliverable
Research tool
M33Planned
Modern analytics / Practitioner
Anomaly detection and monitoring
Outcome
Define seasonality-aware baselines and false-alert costs.
Prerequisites
M23, M24, M25, M29
Datasets
DS025, DS062, DS078
Deliverable
Monitoring note
Wave 06
Publish a decision product
Package reproducible work and translate it into executive action.
3 modules
M34Planned
Publishing / Practitioner
Reproducible notebooks and evidence packages
Outcome
Package code, data manifest, environment, tests, and expected outputs.
Prerequisites
M02, M03, M04, M13
Datasets
DS024, any approved dataset; DS024 is the pilot
Deliverable
Template + guide
M35Planned
Publishing / Reader
AI business value and executive communication
Outcome
Translate an analysis into a board decision without removing uncertainty.
Prerequisites
M05, M10, M13
Datasets
DS063, DS064, DS073, enterprise-safe inputs
Deliverable
CIO crossover
M36Planned
Publishing / Researcher
Capstone: the Schym Signal
Outcome
Publish a versioned signal that can be updated without changing its contract.
Prerequisites
M01, M02, M03, M04, M05, M06, M07, M08, M09, M10, M11, M12, M13, M14, M15, M16, M17, M18, M19, M20, M21, M22, M23, M24, M25, M26, M27, M28, M29, M30, M31, M32, M33, M34, M35
Datasets
choose two to four complementary datasets
Deliverable
Capstone dossier

Follow the evidence, not the software menu

Eight analytical tracks

Question → measurable claim

Provenance, license, and versions

Tidy data, keys, and joins

APIs, JSON, and SDMX

Units, deflators, PPP, and indexes

Missingness, flags, and outliers

Descriptive statistics that answer questions

Distributions, inequality, and concentration

Sampling and survey weights

Uncertainty and bootstrap intervals

Tests, effect sizes, and power

Bayesian updating and Monte Carlo simulation

Chart grammar, integrity, and accessibility

Regression as controlled comparison

Functional form, logs, and interactions

Panel data and fixed effects

Causal diagrams and confounding

Experiments and A/B tests

Difference-in-differences and event studies

Regression discontinuity

Instrumental variables

Synthetic control and counterfactuals

Time series: trend, cycle, and autocorrelation

Seasonality and structural breaks

Forecasting and rolling validation

Geospatial analysis and MAUP

Remote sensing as economic evidence

Networks, flows, and input-output dependencies

Classification, calibration, and asymmetric costs

Clustering and principal components

Text as data

Embeddings, retrieval, and semantic maps

Anomaly detection and monitoring

Reproducible notebooks and evidence packages

AI business value and executive communication

Capstone: the Schym Signal