samiemirza / weather-prediction

This project utilizes univariate and multivariate linear regression, to predict the temperature the next day, analyzing the provided dataset. The file contains a line by line walkthrough of the code, with an explanation of each step, along with ample visual and verbal analysis.

Weather Prediction using Linear Regression

This is an extensive and well-documented implementation of a linear regression model using both univariate and multivariate approaches to predict the next day's maximum temperature based on various features. The provided code covers:

Data Import and Preparation:

Loading the dataset and extracting relevant features and target variables. Splitting the dataset into training and test sets.

Univariate Regression:

Training a model using one feature (Present_Tmax) to predict the next day's temperature. Implementing gradient descent for parameter optimization. Visualizing the results with plots.

Feature Analysis:

Calculating and visualizing the correlation between different features and the target variable (Next_Tmax).

Improving Performance:

Exploring the effect of different features on the model's performance.

Multivariate Regression:

Extending the model to include multiple features. Using gradient descent for parameter optimization. Visualizing the cost over epochs.

Comparison with All Features:

Training a model using all available features. Comparing the final cost values between models trained with a subset of features and all features.

Analysis:

Suggesting experiments with different learning rates and train-test split ratios to observe their impact on model performance.