{"id":31128911,"url":"https://github.com/parseablehq/zero-shot-forecasting","last_synced_at":"2026-02-14T16:31:09.428Z","repository":{"id":313761353,"uuid":"1052563178","full_name":"parseablehq/zero-shot-forecasting","owner":"parseablehq","description":null,"archived":false,"fork":false,"pushed_at":"2025-09-08T08:47:59.000Z","size":11193,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-02-02T06:44:44.897Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/parseablehq.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null},"funding":{"github":["parseable","operator"]}},"created_at":"2025-09-08T08:25:33.000Z","updated_at":"2025-09-08T08:48:02.000Z","dependencies_parsed_at":"2025-09-08T10:41:29.063Z","dependency_job_id":null,"html_url":"https://github.com/parseablehq/zero-shot-forecasting","commit_stats":null,"previous_names":["parseablehq/zero-shot-forecasting"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/parseablehq/zero-shot-forecasting","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/parseablehq%2Fzero-shot-forecasting","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/parseablehq%2Fzero-shot-forecasting/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/parseablehq%2Fzero-shot-forecasting/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/parseablehq%2Fzero-shot-forecasting/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/parseablehq","download_url":"https://codeload.github.com/parseablehq/zero-shot-forecasting/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/parseablehq%2Fzero-shot-forecasting/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29449369,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-14T15:52:44.973Z","status":"ssl_error","status_checked_at":"2026-02-14T15:52:11.208Z","response_time":53,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-09-18T01:20:44.665Z","updated_at":"2026-02-14T16:31:09.423Z","avatar_url":"https://github.com/parseablehq.png","language":null,"funding_links":["https://github.com/sponsors/parseable","https://github.com/sponsors/operator"],"categories":[],"sub_categories":[],"readme":"# Zero‑Shot Forecasting Artifacts\n\n## Contents at a glance\n\n- Forecast plots: `Forecast Plots/`\n- CSV summaries: `CSV Files/`\n\n## Series \u0026 Windows\n\n- Prometheus memory: 10s, 5m, 10m windows\n- OpenSearch CPU: 10s, 5m, 10m windows\n\n## Forecast Horizons\n\n- 64, 128, 256, 336 steps\n\n## Baselines \u0026 Models\n\n- Baselines: \"t\" (current value) and \"t-1\" (last value)\n- Chronos: default quantiles 0.1–0.9 to produce fan charts\n- Toto: varied sampling; (32, 4) highlighted as speed/quality sweet spot\n\n## Metrics\n\n- MASE: point-forecast accuracy versus naïve\n- CRPS: calibrated uncertainty quality for predictive distributions\n\n## Evaluation Protocol\n\n- Rolling forecast origin\n- Fixed final test window\n- Latency recorded per horizon for each method\n\n## Data \u0026 File Structure\n\n- Series naming: `\u003cmetric\u003e_\u003cwindow\u003e_\u003csource\u003e`\n  - Examples: `mem_util_5m_prometheus`, `cpu_util_10s_opensearch`\n- CSV files: `\u003cprediction_length\u003e_\u003cdata_used\u003e.csv`\n  - Example: `512_mem_util_5m_prometheus.csv`\n- Plot folders: `\u003cprediction_length\u003e_\u003cdata_used\u003e/` containing two plots per series\n  - Files: one \"chronos\" plot and one \"toto\" plot\n- Root paths in this repo:\n  - Plots under `Forecast Plots/`\n  - CSVs under `CSV Files/`\n\n## CSV Header Dictionary\n\n- Toto Time: Toto inference time for the horizon (ms or s, as exported)\n- Chronos Time: Chronos inference time for the horizon\n- Toto MASE: MASE for Toto’s point forecast vs t baseline\n- Chronos MASE: MASE for Chronos’s point forecast vs t baseline\n- Toto CRPS: CRPS for Toto’s predictive distribution\n- Chronos CRPS: CRPS for Chronos’s predictive distribution\n- Input Length: Context length used for inference\n\n---\n\nNotes\n\n- Time units in exported CSVs may be milliseconds or seconds depending on source tool.\n- Chronos fan charts are rendered from the 0.1–0.9 quantiles.\n- Toto results reflect different sampling configs; (32, 4) is a good trade‑off.\n\n## Baselines Per Horizon\n\n- Definition: Let the last observed value at the forecast origin be `y_t`, and the previous value be `y_{t-1}`. For a horizon `H` (e.g., 64/128/256/336), the baselines are:\n  - \"t\" baseline (persistence): for all `h = 1..H`, `ŷ_{t+h} = y_t`.\n  - \"t-1\" baseline: for all `h = 1..H`, `ŷ_{t+h} = y_{t-1}`.\n- Usage: These baselines produce a full `H`-step forecast vector by repeating a single observed value across all steps.\n- Metrics: MASE and CRPS are computed per horizon window using these baseline definitions for comparison.\n- Edge cases: If `y_{t-1}` is not available (very early in the series), the \"t-1\" baseline is skipped or marked N/A for that origin.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fparseablehq%2Fzero-shot-forecasting","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fparseablehq%2Fzero-shot-forecasting","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fparseablehq%2Fzero-shot-forecasting/lists"}