Data Mining with R: Go from Beginner to Advanced!

Master input and output in data mining with R by reading data from files or keyboard prompts, manipulating data, exporting results to disk, across six videos and an after-session exercise.

Identify three folders for section two data and file input and output, and use an exercise you can paste into the studio script window with data and scripts from lessons.

Learn how to load data into an R session and workspace, and review core data structures: vectors, matrices, data frames, lists; and use interactive input with scan and readLine.

Learn to use scan() in R to read data from keyboard or disk, specify data types with what, handle numeric and character columns, and convert lists to vectors.

Master how the readline function prompts for user input with the prompt argument in the console and how to assign the input to a variable, with an example in R.

Learn to read text and comma-separated files into R data frames using read.table and read.csv, handling headers, delimiters, and interactive file selection.

Save data frames from your R session to disk with write or write.table, set the working directory, and verify the saved file with file.exists; explore loading the car prestige dataset.

Practice data input and output in R by scanning keyboard data into X and Y, evaluating their mode, reading files with read table, and writing data back as PTSD.

Explore interactive data input with scan and the wrappers read.table and read.csv, including keyboard entry, console focus, and distinguishing headers and delimiters for file reading.

Learn how to read and write data in R using read.csv and read.table, manage headers and row.names, work with data frames, and navigate directories and path syntax.

Master preprocessing and visualization in R by cleaning the workspace, managing memory with gc, and exploring large birth data to plot births by day of week and delivery type.

Learn to preprocess birth data and visualize it with lattice graphics in R, using bar charts, histograms, density plots, and conditioned panels by plurality and delivery type.

Preprocess and visualize birth data using box plots, violin plots, and level plots to reveal distributions, Apgar scores, and gestation relationships.

Analyze alumni donations in R by loading donor data, computing total giving, and visualizing distributions with bar charts and density plots to reveal outliers and fundraising patterns.

Learn to save lattice plots as objects, inspect their structure, and customize box plots to explore alumni donations by year, gender, marital status, and event attendance.

Tabulate giving by major to form a data frame of total giving and frequency. Filter out infrequent donors, then visualize distributions and bar charts by year of graduation.

Explore visualizations of alumni donations using bar plots, mosaic plots, and correlation plots to reveal patterns by graduation year, giving frequency, gender, marital status, and event attendance.

Visualize Virginia's 1941 death rates data with bar charts and Cleveland dot plots, then write your own commands to solve the exercise.

Learn to visualize multivariate Virginia deaths data in R with lattice graphics, converting matrices to data frames, applying conditioning, and comparing rural/urban, male/female groups via dot plots.

Explore decision trees as essential building blocks for data mining and ensembles like random forests, and learn how R packages such as party and rpart support recursive partitioning.

Explore decision trees for classification and regression using the iris dataset, distinguish between CART tasks, and learn a 70/30 train-test split with set.seed for reproducible results.

Explore training a conditional inference tree using the party package in R, predicting iris species from four variables, evaluating with training and test confusion matrix, and interpreting splits and leaves.

Explore building a bushy regression tree for body fat in females with the rpart package, then prune using mean squared error reduction and Gini index as stop criteria.

Explains how to build and prune a body fat decision tree with the rpart package, using minsplit and cp controls, and evaluating pruning via the CP table.

Explore bagging with bootstrap sampling and multiple trees, evaluate via holdout tests and out-of-bag error, then compare to random forests with variable selection and variable importance measures.

Learn to build and compare a decision tree and random forest models on the forensic glass dataset, using rpart and bagging techniques, for classification with training and test splits.

Explore bagging through bootstrap aggregation, growing multiple trees on diverse bootstrap samples to smooth predictions, using out-of-bag estimates and majority voting.

Learn how random forests use random variable subsets at each split to create diverse trees, reducing misclassification, and compare with bagging using out-of-bag estimates and the genie index.

Apply regression and generalized linear modeling to the Hart data set through detailed script-based examples, with exercises in PTF form after reviewing lessons and videos.

Explore ordinary least squares regression and fitted values, with residuals and key assumptions. See how generalized linear modeling extends linear modeling for non-normal data, using R on a CPI example.

continue regression example by fitting a linear model to CPI with year and quarter, interpreting coefficients, intercept, and r-squared, and predicting CPI for future quarters.

Explore regression diagnostics and predictions by examining residuals, standardized residuals, q-q plots, leverage and Cook's distance, plus 3d scatterplots and CPI forecasts for 2011.

Explore linear regression and GLMs, including logistic regression with a logit link, and learn how error structures and non-normal responses influence model choice and interpretation.

Explore generalized linear modeling, transforming the linear predictor with link functions for binomial data, including logistic regression using the logit and proportions example.

Model heart data as a proportion using a binomial glm in R, with counts of heart attacks and non-attacks, and compare models via residual deviance, diagnostic plots, and chi-square tests.

Explore Poisson GLM modeling of AIDS case counts over time, diagnose fit with plots, and improve the model by adding a quadratic term, then compare via deviance and ANOVA.

Fit three linear models, including first and second order terms, to identify the best fit, then run two separate GLMs to apply recent modeling concepts.

Review session 6 materials for the first cluster analysis, including two folders with regression exercise solutions (pdf) and scripts with answers, plus end-of-session exercises on residuals from path models.

Explore regression and GLM concepts in data mining with R by comparing linear, quadratic, and intercept-free models of braking distance from speed using the cars data.

Explore regression and generalized linear models to model counts, comparing linear and quadratic terms, addressing error distributions, poisson glm, and model selection via deviance and aic.

Explore cluster analysis with k-means on the iris data, comparing three versus two clusters while examining centroids, distance measures, and within-cluster sums of squares to reveal patterns.

Explore k-means clustering on exoplanet data, using mass, periodicity, and eccentricity, visualized in a three-dimensional scatterplot with range normalization to form and compare three candidate clusters.

Explore how k-medoids assigns each iris observation to the nearest cluster center using distance, via pam and pamk, and evaluate two- and three-cluster solutions with silhouette plots.

Explore hierarchical clustering on iris, compare single, complete, and average approaches, and cut the tree to form three groups using average Euclidean distance.

Apply hierarchical clustering to 45 Romano British pottery shards to infer kiln membership from chemical traces, using Euclidean distance and single, complete, and average linkage, with level plots.

Explore density-based and hierarchical agglomerative clustering materials for section 7, including Becher 2013, Kamins cluster analysis, Tamm residuals, MDA multivariate analysis, f pci-e, scripts, and slides.

Explore density-based and agglomerative clustering, using residuals from latent variable path models to reveal latent heterogeneity; apply k-means based pos clustering for segmentation.

Explore density-based clustering with dbscan in R using the fpc package, focusing on reachability distance and min points, identifying noise, and visualizing results on the iris dataset.

Explore agglomerative hierarchical clustering on a body measurement dataset (hips, chest, waist) of 20 observations, comparing single, complete, and average linkage to reveal gender-based groups.

Apply hierarchical clustering to body measurement data, compute Euclidean distance among waist, hips, and chest, compare seven linkage methods, and use dendrograms and principal components to distinguish men from women.

Explore clustering techniques in data mining with R, comparing hierarchical and k-means, using scaling and complete linkage on jet fighters to classify carrier landing ability via principal components.

Explore section 8's final cluster analysis examples, tackle the residual analysis exercise, and compare standardization by standard deviation versus by range using crime rate data.

Explore k-means clustering in data mining with R, compare it with hierarchical and model-based approaches, and learn centroid-based assignment that minimizes the sum of squared residuals.

Explore k-means clustering on U.S. crime arrest rates across 7 variables, address within-group sums of squares, scaling, and outliers, then validate splits with scatterplot matrices and principal components.

Identify outliers in crime rate data and use range normalization to standardize variables before applying k-means clustering, then assess elbow plots and visualize two-group separation on principal components.

Explore a data mining workflow with R: expand a six-question Tompson questionnaire across 28 countries, then apply model-based clustering to reveal country patterns in medical disclosure practices.

Explore graphical approaches to cluster analysis, using neighborhood plots and stripes plots to visualize centroid distances, cluster overlap, and structure from toy multivariate data.

Explore how box plots and the underlying stats identify outliers beyond whiskers, explain univariate and multivariate outliers with X and Y, and visualize distributions with density plots.

Explore univariate and multivariate outliers using box plots, intersection and union strategies, local outlier factor, and k-means clustering on iris data with visualizations.

Learn k-means clustering in R on crime rate data, standardizing variables with z-scores, and explore how random initial partitions and minimization of within-group sums of squares form clusters.

Explore scatterplot matrices, scaling by centering and standardization, and PCA to analyze crime data in R. Identify outliers like DC and compare covariance and correlation matrices.

Present a final case study of a software project in R using k-means clustering to illustrate data mining, with scripts and slides on POS prediction oriented segmentation.

Explore k-means clustering for identifying unobserved heterogeneity in structural equation models. Implement group-specific PLS path modeling in R, including data splitting and evaluating residuals and R-squared.

Apply k-means style reallocation to PLS path models by minimizing residuals and maximizing group-wise R-squared, re-evaluating all models after each move to improve homogeneity.

Explore a three-way multi-group analysis using permutation tests to compare path coefficients, loadings, weights, and R-squared across groups, with pairwise comparisons and bootstrapped differences.

Learn to perform k-means clustering on TAM residuals in R, build and load an R package, run the algorithm on large datasets, and interpret R-squared and the resulting group memberships.