Solar Power Prediction Eval

---
name: solar-power-prediction-eval
description: Evaluates the predictive accuracy of ensemble machine learning models for forecasting solar power generation using meteorological parameters. It probes regression performance under varying feature sets and ensemble aggregation strategies. Use when the user wants to benchmark on SRRA dataset, or asks about evaluating this task. Reports RMSE.
metadata:
  skill_kind: dataset_eval
  source_arxiv: 2301.10159
  bibtex_key: chakraborty2023ensemble
  confidence: high
---

# solar-power-prediction-eval

> Computational Solar Energy -- Ensemble Learning Methods for Prediction of Solar Power Generation based on Meteorological Parameters in Eastern India — Chakraborty et al. (2023) (arXiv:2301.10159, 2023)

## What this evaluates

Evaluates the predictive accuracy of ensemble machine learning models for forecasting solar power generation using meteorological parameters. It probes regression performance under varying feature sets and ensemble aggregation strategies.

## Datasets

- **SRRA dataset** — total ?; splits: train (-1), test (-1)

## Metrics

- `RMSE` **(primary)** — range: other
  - RMSE = sqrt(sum((x_i - x_hat_i)^2) / N). Measures the square root of the average squared difference between predicted and actual power generation. The paper uses x_i notation for targets.
- `R-Square score (R^2)` — range: [0, 1]
  - R^2 = SS_regression / SS_total. Represents the proportion of variance in the target variable explained by the regression model.

## Input / output format

**Input**: Meteorological parameters (features) corresponding to a specific time interval.

**Output**: Continuous predicted solar power generation value.

## Scoring recipe

```python
def rmse(y_true, y_pred):
    return np.sqrt(np.mean((y_true - y_pred) ** 2))

def r2(y_true, y_pred):
    ss_res = np.sum((y_true - y_pred) ** 2)
    ss_tot = np.sum((y_true - np.mean(y_true)) ** 2)
    return 1 - (ss_res / ss_tot)
```

## Common pitfalls

- The dataset is referred to as 'SRRA dataset' but no download link, repository, or exact size is provided, making direct replication difficult.
- The RMSE formula in the paper uses x_i and x_hat_i notation instead of standard y_i/y_hat_i, which may cause confusion when implementing the metric.
- Hyperparameter tuning details are listed as final values only, without specifying the search space or optimization method used.

## Evidence (verbatim from paper)

> To evaluate model's performance, some evaluation metrics have been defined which are root mean squared error (RMSE) and R-Square score $(R^2)$ . The prediction framework has been represented in Figure 12. $$ \mathrm {R M S E} = \frac {\sqrt {\sum_ {n = 1} ^ {N} \left(x _ {i} - \hat {x} _ {i} \right) ^ {2}}}{N} \tag {10} $$ $$ R ^ {2} = \frac {S S _ {\text {r e g r e s s i o n}}}{S S _ {\text {t o t a l}}} \tag {11} $$

## Citation

```bibtex
@misc{chakraborty2023ensemble,
  title={Computational Solar Energy -- Ensemble Learning Methods for Prediction of Solar Power Generation based on Meteorological Parameters in Eastern India},
  author={Chakraborty et al. (2023)},
  year={2023},
  note={arXiv:2301.10159}
}
```

- arXiv: 2301.10159