internship-lanes / README.md
asca-mirza-7's picture
lane example cuts (June 2026) + provenance card
0304775 verified
|
Raw
History Blame Contribute Delete
2.99 kB
metadata
license: other
language:
  - en
tags:
  - seo
  - content-performance
  - tabular
  - education
  - flyrank-internship
pretty_name: FlyRank Internship  Lane Datasets (Pseudonymized, Gated)
size_categories:
  - 1M<n<10M
extra_gated_prompt: >-
  By requesting access you agree to the FlyRank Internship Data Use Terms:
  anonymized research and education use only; no attempt to re-identify clients,
  domains, queries, keywords, or content; no redistribution of the raw data; and
  no client-identifying data in any public output (case study, repo, chart, or
  demo).
extra_gated_fields:
  Name: text
  Email: text
  Affiliation or cohort: text
  I agree to the FlyRank data-use terms: checkbox
configs:
  - config_name: ranking_lifecycle
    data_files: default_lanes/ranking_lifecycle.parquet
  - config_name: decline_recovery
    data_files: default_lanes/decline_recovery.parquet
  - config_name: engagement_fix
    data_files: default_lanes/engagement_fix.parquet
  - config_name: ai_opportunity
    data_files: default_lanes/ai_opportunity.parquet
  - config_name: cannibalization_restricted
    data_files: restricted_lanes/cannibalization_restricted.parquet
  - config_name: semantic_clustering_restricted
    data_files: restricted_lanes/semantic_clustering_restricted.parquet

FlyRank Internship — Lane Datasets (Pseudonymized)

Six task-focused, pseudonymized lanes derived from the FlyRank warehouse. Salted/namespaced hashed keys join across lanes and the warehouse release.

Provenance: these are fixed example cuts prepared in June 2026 (90-day windows as of their build date). They exist to show the shape of lane work — the warehouse release (v2 full history) is the source of truth, and capstone-grade features should be engineered from it directly (see starter notebook 03).

Default lanes (intern-safe)

Lane Rows Typical task
ranking_lifecycle 205,749 ranking lifecycle, archetypes, baseline scoring
decline_recovery 56,253 decline/recovery classification (define a forward label)
engagement_fix 33,202 CTR / engagement opportunity analysis
ai_opportunity 11,571 AI-referral pattern analysis (positive-only — not a clean binary classifier)

Restricted-origin lanes (raw query/URL origin → keep separate)

Lane Rows Typical task
cannibalization_restricted 712,454 cannibalization severity / action ranking
semantic_clustering_restricted 350,707 structured clustering / archetypes

Rules

Load a lane with datasets:

from datasets import load_dataset
ds = load_dataset("FlyRank/internship-lanes", "ranking_lifecycle")

Product flags / scores (health_score, trend_*, recommended_action, action_type) are leakage-risk context, not default features. Define your own labels, features, splits, and a leakage-safe prediction point. Public outputs must be anonymized/aggregate only.