Practical Recommender Systems For Business Applications in R

Learn to read csv and excel data in R studio, using read functions with headers, inspect with head, and import datasets from csv, text, or excel formats into data frames.

Read data from online HTML tables, like the 2016 Summer Olympics medal table on Wikipedia, into R using getURL and readHTMLTable, then extract and process the table for analysis.

Read data from html tables on wikipedia pages using the Harvest library, handling null tables and export details to extract the 2016 Olympics medal table and UK World Heritage sites.

Explore data cleaning in R by handling missing data with complete cases, mean substitution, and imputation using mice and predictive mean mapping, illustrated on the Boston and air quality datasets.

Read a csv file, clean GDP per capita by removing dollar signs and commas, convert to numeric, and rename population in millions to population and bio capacity to bio capacity.

Learn to pre-process data in R using the pipe operator to chain functions with the player package. Practice selecting, dropping, and filtering the iris dataset by conditional selections.

Master data summarizing with dplyr in r by selecting, filtering, mutating, and summarizing air quality data with ozone, solar, wind, temp, day, and month, then computing per-month means.

Use the dplyr pipe operator to chain functions like select, group by, and summarize, computing mean values and applying filters to reveal monthly data patterns.

Learn to join data frames in R using dplyr, including left, inner, and full joins, summarize plays, and link to song data to reveal song title and artist.

Explore reshaping user–song data from long to wide using the spread function, turning each song into its own column for a top 1000 songs dataset.

Explore data visualizations for recommender systems using the recommender lab package and built-in movie data, and interpret rating frequencies to understand data before modeling.

Explore principal component analysis, an unsupervised dimensionality reduction technique that converts correlated predictors into uncorrelated linear components, revealing maximum variation and guiding pattern discovery.

Learn to implement principal component analysis in R to reduce dimensionality, center and scale data, and interpret variance explained and loadings through practical visualization.

Introduce singular value decomposition (SVD) as a three-matrix decomposition of a matrix, and describe its use for feature extraction, decomposition, and image reconstruction in recommender systems.

Implement singular value decomposition in R by scaling the data, computing the svd, and using the singular values with left and right vectors to gauge the first component's 85% variance.

Explore unsupervised learning theory through remote sensing imagery, showing how unlabeled spectral band data cluster into land-use patterns via unsupervised classification.

Explore unsupervised clustering with k-means, which partitions data into k clusters by assigning observations to the nearest centroid using Euclidean distance, iteratively refining centers.

Learn how to implement k-means for unsupervised clustering. Choose the number of clusters, initialize centers, assign by Euclidean distance, minimize within-cluster dispersion, and assess results with the iris data.

Learn the theory and applications of supervised classification in remote sensing, using training sites to define class means, variances, and spectral similarity for pixel assignment.

Implement cosine similarity in R by applying a cosine function to a matrix of vectors, enabling recommender systems to identify similarity between users and products.

Explore practical recommender system types, including content-based, collaborative, and hybrid filtering, with memory-based and model-based approaches, item-based and user-based strategies, plus clustering, neural networks, spatial networks, and association techniques.

Explore the recommenderlab package, inspect the recommender registry of models, examine a rating matrix, and review user-based collaborative filtering and gradient-descent based approaches along with their typical parameters.

Format your data to recommenderlab specifications, preprocess by trimming to a subset (about 70%), and convert the data frame into a real rating matrix ready for modeling.

Explore building a cosine similarity based recommender engine in R, using a real rating matrix, train-test splits, and evaluation to predict top ten items.

Explore building a recommender system with the recommender lab by preparing a real rating matrix, applying cross-validation, comparing models (cosine-based user-based collaborative filtering), and generating personalized predictions.

Explore collaborative filtering using cosine similarity to identify similar users by ratings and derive item recommendations from a pre-processed user-item rating matrix.

Use clustering analysis on books based on characters and ratings, leveraging a Goa distance metric to identify similar titles and generate recommendations.

Identify top reader preferences by building a dot matrix from book ratings and categories, using a similarity and dissimilarity metric across user IDs to reveal individual preferences.

Build an item-based recommender from a user-item ratings matrix using a cost similarity approach to generate top book recommendations.

Learn how to run R in Google Colab by mounting Google Drive, loading the R extension, and executing R code with installed packages and data access.

Word clouds visualize textual data by sizing frequent words to reveal a topic, removing common words. Build them from sources like tweets or Wikipedia, using Python or workflow generators.