rosannagamal/SolarPanelPowerOutputPrediction

This project aims to develop a robust machine learning model to predict the solar power output of panels in Egypt, taking into account recent climate changes and their impact on weather conditions. # Solar Panel Power Output Prediction in Egypt This project aims to develop a robust machine learning model to predict the solar power output of panels in Egypt, taking into account recent climate changes and their impact on weather conditions. ## Problem Statement Solar energy production in Egypt is influenced by changing weather patterns, especially due to climate change. Existing models often fail to reflect these recent shifts. This project builds a model that: * Predicts daily solar panel power output. * Incorporates multiple weather parameters. * Compares different regression algorithms for optimal accuracy. ## Dataset * **Source:** Weather data collected from Aswan, Egypt. * **Records:** 398 daily observations. * **Features:** * `Date` * `AvgTemp` (Fahrenheit) * `AverageDew` * `Humidity` * `Wind` * `Pressure` * `Solar(PV)` – Target variable (power output in watts) ## Data Preprocessing ### Steps Taken: 1. **Date Parsing:** Converted `Date` from object to datetime. 2. **Missing Values:** Linear interpolation applied to missing `AvgTemp`. 3. **Outlier Removal:** * Used **IQR** and **MAD** techniques. * Replaced negative solar values with mean. 4. **Feature Scaling:** Used `StandardScaler` for normalization. ## Exploratory Data Analysis * **Correlation Matrix:** Identified strong positive correlations between `Solar(PV)` and: * `Humidity` (0.72) * `AverageDew` (0.66) * `Wind` (0.34) * **Distributions & Outliers:** Visualized via histograms and boxplots. * **Temporal Trends:** Time series plots revealed seasonal and daily patterns. ## Feature Selection Used **Pearson's Correlation** to select features most correlated with solar output: * Included: `Humidity`, `AverageDew`, `Wind` * Excluded: `Pressure`, `AvgTemp` (weak/negative correlations) ## Models Evaluated | Model Type | MAE | RMSE | R² | Adj. R² | Accuracy | | ------------------------- | ---- | ---- | -------- | -------- | -------- | | Linear Regression (No FS) | 0.3 …