The Complete Supervised Machine Learning Models in R

Learn how the simple linear regression model builds on the mean model to predict land prices from area, visualize data with a mean line, and evaluate residuals and SSR.

Define the simple linear regression equation y = mx + z, where y is the dependent variable and x is the independent variable, to predict discounts from bill amount.

Derive the simple linear regression equation by computing the slope and intercept from deviations around means, plot the regression line, and assess fit with residuals and R-squared.

Use bill amount as the independent variable and discount as the dependent variable to train and test a simple linear regression model and predict discounts from bill amounts.

Read the bill discount csv, define bill amount as the independent variable and discount as the dependent variable, and split data with a seed for training and testing in R.

Train a simple linear regression model in R using lm, evaluate with training and testing data, and interpret R-squared and residuals to predict future discounts.

Visualize simple linear regression in R by plotting training and testing data with ggplot2, and drawing regression lines to predict discounts from bill amounts.

Compare training and testing R squared values to assess model performance, showing 0.975 training and 0.984 testing, indicating no overfitting, as we prepare to template a simple linear regression model.

Build a simple linear regression template in R to predict discounts from bill amounts using the predict function, handling single and multiple bill inputs with data frames and vectors.

Create a template for simple linear regression in R, featuring a graphical user interface input for bill amount and a button to predict and display the discount.

Explore how multiple linear regression extends simple regression to predict a dependent variable from many independent variables, including the equation and coefficient interpretation while selecting relevant predictors to avoid overfitting.

Explore the concepts of underfitting and overfitting, learn how model diagnosis helps identify weak inputs, and understand correlation as the degree of relationship between variables, with examples from regression.

Explore p values and null and alternative hypotheses, and understand adjusted R-squared to diagnose and improve multiple linear regression models in R.

Learn how to diagnose and simplify models to improve accuracy in multiple linear regression by inspecting data, removing nonsense variables, evaluating correlations, and avoiding multicollinearity.

Convert categorical data to numerical form using dummy variables to enable linear regression, illustrate with city and country examples, and show how to handle two or more categories.

Explore forward selection and backward elimination for feature selection in linear regression, using adjusted R square and P value to identify impactful independent variables against the dependent variable.

Split the data into training and testing sets to train a multiple linear regression model using area, distance, and crime rate to predict price; build a template for predictions.

Read the land price data from a csv, implement multiple linear regression in R with area, distance, and crime rate predicting price, and split data 60/40 for training and testing.

Train a multiple linear regression model in R using lm to express price as a function of area, distance, and crime rate, and assess p-values and R-squared to refine model.

Train and test a multiple linear regression model in R using area, distance, and crime rate to predict land prices and evaluate predictions with the predict function.

visualize a four-variable multiple linear regression with three independent and one dependent variable, prioritizing area over distance and excluding crime rate due to nonlinear relation; prepare for the next assignment.

Use the dummies package in R to create city dummies, then train a multiple linear regression with area, distance, crime rate, and Interlaken and Geneva.

Builds a template for a multiple linear regression model in R, using area, distance, and crime rate to predict land price with the predict function and a new data frame.

Learn to build a multiple linear regression template in R, loading data, splitting train and test sets, and predicting land price from area, distance, and crime rate.

Master polynomial regression as a curvilinear extension of multiple linear regression, adding higher-degree terms to fit data and improve predictions. Visualize with scatter plots and evaluate using R square.

Visualize data for polynomial regression in R using a ten-observation data set, with angle as the independent variable and distance as the dependent variable, via a scatter plot.

Train a simple linear regression model in R to predict distance from angle, then plot and build polynomial models up to degree 2 and 3, and compare fits.

Explore how to compute and compare R squared for linear and polynomial regression in R, with degrees two and three, and learn how overfitting affects test performance.

Build a polynomial regression template in R to predict ball distance from angle and angle degree two, using predict with poly and data frames, including single and multiple angle inputs.

Build a polynomial regression template in R (degree two) to predict distance from an input angle, using a distance predictor interface and the predict function.

Explore bias and variance in regression and how regularization reduces overfitting by shrinking coefficients. Understand ridge regression, lasso, and elastic net, and how penalties balance training and testing performance.

Explore ridge regression, a regularization technique that reduces model complexity by adding a penalty on coefficients. Learn how lambda shrinks coefficients to curb overfitting and improve performance on unseen data.

Use a height–weight dataset to illustrate ridge regression, with height as the independent variable and weight as the dependent variable, including example values. Access datasets and code files by email.

Learn to implement ridge regression in R using a single predictor, train with a 30/70 split, and evaluate R-squared on training and testing data while visualizing the regression line.

Train ridge regression in R with glmnet to bias the regression line and reduce overfitting, tuning alpha (0) and lambda while comparing training and testing R-squared.

Build a ridge regression template in R with glmnet, converting a data frame to a matrix and predicting weight from height using a constant one.

Build and customize a ridge regression template in R by adapting the simple linear regression code, reading data, splitting training and testing sets, and predicting weight with glmnet.

Explore lasso regression, a regularization technique that shrinks coefficients and can zero out variables for feature selection, reducing complexity and overfitting, unlike ridge.

Explore a height–weight dataset (height in cm as the independent variable, weight in kg as the dependent variable) for lasso regression model, with datasets and code files available via email.