# KEV0143/Comparative-analysis-of-hourly-load-forecasting-using-PatchTST-TFT-NHiTS-and-CatBoost

A comprehensive time-series benchmark evaluating state-of-the-art deep learning architectures (PatchTST, TFT, N-HiTS) against traditional gradient boosting (CatBoost) for accurate 24-hour load prediction.

Repository: https://github.com/KEV0143/Comparative-analysis-of-hourly-load-forecasting-using-PatchTST-TFT-NHiTS-and-CatBoost
Canonical: https://ross.abutalabs.com/products/comparative-analysis-of-hourly-load-forecasting-using-patchtst-tft-nhits-and-catboost
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-07-25T06:32:07+00:00

## Health v2 (maintenance only)
Score: 63/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 94, release rhythm 46, longevity 25
- inputs: {"age_days": 352, "days_push": 39, "days_rel": 147, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1676, forks 0 (observed 2026-08-28T04:05:20.365102+00:00)

## What it is
A Python benchmark project comparing deep learning time-series architectures (PatchTST, Temporal Fusion Transformer, N-HiTS) against CatBoost gradient boosting for 24-hour electricity load forecasting. It includes data preprocessing, model training, validation, and result visualization for energy market decision support.

## Use cases
- compare deep learning vs gradient boosting for hourly load forecasting
- benchmark PatchTST against TFT and N-HiTS on time-series data
- predict electricity demand 24 hours ahead
- evaluate CatBoost for time-series regression
- find the best model architecture for energy load prediction
- learn how to run forecasting experiments with multiple models

## When to choose
- you need a ready-made comparison of modern time-series architectures for load forecasting
- you work in energy analytics and need hourly demand predictions
- you want a reference implementation of PatchTST, TFT, N-HiTS, and CatBoost in one pipeline

## When to avoid
- you need a production forecasting service rather than a research benchmark
- your forecasting problem is unrelated to hourly energy load
- you need a maintained library with an API instead of an analysis codebase

## Facets
- artifact type: learning-resource
- maturity: active
- function: machine-learning, benchmarking, data-science, data-visualization, etl
- domain: machine-learning, data-science, time-series, energy, analytics
- platform: python, windows
- tags: time-series-forecasting, load-forecasting, patchtst, temporal-fusion-transformer, n-hits, catboost, deep-learning, energy-markets, benchmark, linux

## Member repositories
- KEV0143/Comparative-analysis-of-hourly-load-forecasting-using-PatchTST-TFT-NHiTS-and-CatBoost (main) score 63

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:20.365102+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:41:38.998954+00:00, confidence not recorded.
  - readme: https://github.com/KEV0143/Comparative-analysis-of-hourly-load-forecasting-using-PatchTST-TFT-NHiTS-and-CatBoost (fetched 2026-08-28T04:05:20.365102+00:00, sha 181a3e124f53)
- Data as of 2026-08-30T08:39:29.467469+00:00.