Machine Learning Using R: Learn Data Science Using R

Identify the essential system requirements for installing R and R Studio: a 64-bit OS, R version 3.0.1 or higher, and 8 to 16 GB RAM.

Explore how machine learning powers food and restaurant recommendations, predicts drug interactions and side effects, and forecasts traffic patterns for real-world decision making.

Explore the full machine learning workflow from defining the question and collecting data to feature selection, training, evaluation, and building a final golden model using a COVID-19 patient dataset.

Explore supervised learning foundations, comparing classification and regression, with real-world applications such as object identification, fraud detection, speech recognition, spam detection, and medical diagnosis.

Understand the intercept as the expected value of the dependent variable when all independent variables equal zero, illustrated with single-variable and multi-variable examples such as height, weight, and house price.

Understand how multiple regression uses several predictors, like house price, to predict a dependent variable, and how the correlation coefficient (multiple r) indicates relationship strength.

Learn to manually calculate linear regression parameters beta0 and beta1 to predict weight from height, train with a small data set, derive the regression line, and tune for accuracy.

Master calculating linear regression parameters in Excel, using height as X and weight as Y, by computing means, sums, and the beta0 and beta1 coefficients to predict weight.

Learn how to interpret linear regression outputs in Excel, including beta one interpretation, standard error, p-values, and 95 percent confidence intervals, with height and weight as example predictors.

Explain the nearest neighbor approach as a supervised learning method for classification and regression. Classify a new data point by comparing its features with neighboring data points.

Explore kNN applications in agriculture, crop protection and yield prediction, groundwater level forecasting, and in personalized recommendations and facial recognition systems.

Explore how Euclidean distance measures the similarity between two data points and guides the classification of a new point by identifying which class its distance is closest to.

Apply the k-nearest neighbors approach with Euclidean distance to predict weight from age and height, and classify the sixth observation as overweight or normal weight using its nearest neighbors.

Explain how the confusion matrix evaluates model performance by comparing actual versus predicted labels in a two-class problem, outlining true/false positives and negatives.

Analyze true positives by comparing actual and predicted class labels, using a threshold of 0.5 to convert probabilities into binary predictions.

Identify true negatives by comparing actual and predicted negative classifications in the data. Note two observations where both actual and predicted values are negative, illustrating true negative cases.

Analyze how false positive cases arise in binary classification by comparing actual negative data points to predicted positive outcomes, and learn how misclassifications occur.

Explore other measures of KNN model accuracy, including sensitivity, specificity (true negative rate), detection rate, detection prevalence, and balanced accuracy, to evaluate prediction performance.

Apply the k-nearest neighbors algorithm in R to classify admissions by GRE, GPA, and college rank. Preprocess admit as yes/no, use an 80/20 train-test split, and evaluate variable importance.