透過分析決策學習 ENA。

33 節課

頁 3 / 6

FirstGame mean network and ENA points. Original figure from the ENA and ONA tutorial by Tan and colleagues (2024).

建模中階35 分鐘2026年7月30日

Use Means Rotation for a Prespecified Two-Group Question

Use the shared synthetic design-talk dataset to practise deciding when the first ENA axis should align with the difference between exactly two prespecified group means, document the decision, test a defensible alternative, and keep every interpretation tied to source evidence.

經審核的教程正文目前以英文提供。

means rotationtwo-group contrastMR1

開啟教程

Tan et al. (2024), figure 5 · CC BY 4.0

Mean network for all units in the expert model. Original figure from the ENA and ONA tutorial by Tan and colleagues (2024).

建模中階35 分鐘2026年7月31日

ENA 學院課程 14

Interpret SVD Dimensions and Co-registration

Use the shared synthetic design-talk dataset to practise understanding what SVD axes summarize and whether network node positions are aligned well enough with projected unit coordinates for interpretation, document the decision, test a defensible alternative, and keep every interpretation tied to source evidence.

經審核的教程正文目前以英文提供。

singular value decompositionco-registrationvariance explained

開啟教程

Tan et al. (2024), figure 27 · CC BY 4.0

Constructing an ENA model in rENA. Original figure from the ENA and ONA tutorial by Tan and colleagues (2024).

建模中階35 分鐘2026年8月1日

ENA 學院課程 15

Reproduce an ENA Model Across Implementations

Use the shared synthetic design-talk dataset to practise checking whether two ENA implementations produce equivalent accumulated connections, normalized vectors, projections, and interpretations under matched settings, document the decision, test a defensible alternative, and keep every interpretation tied to source evidence.

經審核的教程正文目前以英文提供。

cross-implementation parityrENAjENA

開啟教程

Tan et al. (2024), tutorial example · CC BY 4.0

Novice networks projected into the expert space. Original figure from the ENA and ONA tutorial by Tan and colleagues (2024).

研究設計中階35 分鐘2026年8月2日

ENA 學院課程 16

Run Window and Segmentation Sensitivity as a Model Grid

Use the shared synthetic design-talk dataset to practise separating two coupled context decisions by evaluating a small prespecified grid of segmentation and moving-window alternatives, document the decision, test a defensible alternative, and keep every interpretation tied to source evidence.

經審核的教程正文目前以英文提供。

sensitivity gridsegmentationwindow size

開啟教程

Tan et al. (2024), figure 28 · CC BY 4.0

Difference between the two mean ENA networks. Original figure from the ENA and ONA tutorial by Tan and colleagues (2024).

詮釋中階35 分鐘2026年8月3日

ENA 學院課程 17

Read ENA Subtraction Networks Without Overclaiming

Use the shared synthetic design-talk dataset to practise interpreting a difference graph as edge-wise subtraction between comparable networks in one shared analytic space, document the decision, test a defensible alternative, and keep every interpretation tied to source evidence.

經審核的教程正文目前以英文提供。

subtraction networkdifference graphshared node positions

開啟教程

Tan et al. (2024), figure 3 · CC BY 4.0

Source records highlighted within a stanza window. Original figure from the ENA and ONA tutorial by Tan and colleagues (2024).

詮釋入門25 分鐘2026年8月4日

ENA 學院課程 18

Trace Every Interpreted Edge Back to Excerpts

Use the shared synthetic design-talk dataset to practise building a reproducible route from an interpreted network edge to the units, rows, and source excerpts that contributed to it, document the decision, test a defensible alternative, and keep every interpretation tied to source evidence.

經審核的教程正文目前以英文提供。

excerpt traceabilityedge provenanceinterpretive loop

開啟教程

Tan et al. (2024), figure 13 · CC BY 4.0