P160B124_Machine_Learning_Methods

Machine Learning Methods, P160B124

Project Title: Predictive Analysis of Bike Sharing Trends using Linear Regression

Introduction

This project aims to apply linear regression to predict the hourly count of bike rentals over a span of one year. The dataset provided contains hourly rental data, with additional features such as the date, hour, temperature, and humidity, amongst others. The primary goal is to train a linear regression model on the training dataset to learn the associated coefficients and assess the model's accuracy using a separate testing dataset. The datasets are split such that the training dataset includes data from the first 25 days of each month, while the testing dataset includes the remaining days of each month.

Dataset

The dataset is divided into two parts:

Training Dataset: For learning linear regression coefficients.
Testing Dataset: For assessing the accuracy of the trained model.

Features

Date
Hour
Temperature
Humidity
...

Dependent Variable

Rented Bike Count

Tasks

Task 1: Data Loading and Exploration

Load the training and testing datasets (train_bike.csv and test_bike.csv).
Explore the datasets to understand the number of features, types of these features, and the number of observations in both datasets.

Task 2: Feature Visualization

Select two features (excluding Date) and plot them against the Rented Bike Count.
Discuss the potential usefulness of these features based on visual interpretation.

Task 3: Correlation Analysis

Compute the matrix of correlations between numerical variables using the cor function.
Visualize the correlation matrix using the ggcorrplot library.
Select a pair of features with the highest correlation and discuss the impact of removing one feature from the dataset.

Task 4: Linear Regression Model Training

Train a linear regression model using the lm function with Rented Bike Count as the response and all other variables (excluding Date) as predictors.
Predict the count of rented bikes for the testing dataset and plot the predictions to evaluate the model.

Task 5: Variable Transformation

Apply a log transformation to the response variable to address any identified issues.
Retrain the linear regression model and evaluate the model's performance using the R² measure of fit.

Task 6: Interaction Effect Analysis

Explore interaction effects between different features using the * symbol in the model formula.
Evaluate the usefulness of interaction effects by comparing models with and without interaction terms.

Task 7: Feature Transformation Exploration

Explore different transformations (e.g., log, square root, square) for at least four numerical features.
Evaluate the impact of these transformations on the model's performance.

airidas23/P160B124_Machine_Learning_Methods