Solar Energy Production Prediction

🌞 Solar Energy Production Prediction

Problem Description

The goal is to build a regression model that predicts the daily electrical energy production (kWh) of a solar panel, based on meteorological conditions and installation characteristics.

Each sample represents a production day and is characterized by several numerical attributes, such as light intensity, air temperature, wind speed, and others.

The target label (energy_output) represents the total energy generated on that day.

This problem belongs to the univariate regression category.

---

🔹 Features

• solar_irradiance – average solar radiation (W/m²)
• temperature – average air temperature (°C)
• humidity – relative humidity (%)
• wind_speed – average wind speed (m/s)
• cloud_cover – average cloud cover (%)
• panel_angle – panel tilt angle (°)
• panel_efficiency – panel efficiency (%)

---

📘 Input File Structure

train.csv

Contains all feature columns plus the energy_output column, which represents the target value.

Example:

SampleID	solar_irradiance	temperature	humidity	wind_speed	cloud_cover	panel_angle	panel_efficiency	energy_output
1	750.5	25.2	40.0	3.5	10	30	18.5	42.3
2	610.0	22.1	55.0	2.0	50	25	17.0	28.7

test.csv

Contains the same columns as train.csv, but without energy_output, and includes SampleID.

---

📤 Submission

The output file (submission.csv) must contain exactly two columns:

• SampleID
• energy_output – the value predicted by the model (float, with 2 decimal places)

Example:

SampleID	energy_output
1	41.75
2	29.10
3	35.80

---

⚙️ Evaluation

Model evaluation will be performed using Root Mean Squared Error (RMSE):

where N is the number of examples in the test set, y_i is the real value and y^_i is the value predicted by the model.

The final score will be scaled between 0 and 100, so that low RMSE leads to high score.

---

📊 Source

The data used for this problem is synthetically generated.