End to End Data Science Practicum with Knime

Explore data science project management with SEMMA, KDD, and CRISP-DM, from data selection to modeling and evaluation, within a KNIME-driven end-to-end practicum.

Download Knime from knime.com using the installer for your operating system, then drag the Knime app into the applications folder; allow about 1.35 gigabytes of disk space.

Explore the Knime environment and screens by launching Knime, setting the workspace, and navigating the explorer, node repository, console, and outline as you create your first workflow.

Build your first Knime workflow by creating a data table, configuring nodes, and executing data flow from source to destination, then visualize results with a color manager and scatter plot.

Explore categoric and numerical data in Knime, and distinguish nominal, ordinal, interval, and ratio types to guide data understanding, preparation, and modeling.

Explore data access in knime by using file reader, excel reader, and table creator to load and create datasets, configure headers, and prepare data for visualization and preprocessing.

Explore basic data visualization in Knime to discover data behavior and quality, using scatter plots, line plots, and a color manager to analyze height, weight, gender, and stock data.

Learn row filtering and handling missing values, anomalies, and noisy data in Knime to clean datasets before modeling, using range checks and practical data preprocessing techniques.

Explore advanced filtering in Knime using a rule based row filter to combine height, weight, and other rules with and, or, and parentheses for precise data preprocessing.

Explore how to filter data columns using the column filter node in knime, including include/exclude options, type-based selections, and parallel or serial filtering.

Learn to concatenate two data sources in Knime with the concatenate node, matching headers, handling case sensitivity, and managing missing values when combining tables.

Learn to merge data sources in Knime using the joiner to perform inner, left outer, right outer, and full outer joins on IDs to build a unified table.

Learn how to use Knime's group by and aggregation to summarize data, compute means, medians, minimums and maximums, and manage lists and missing values for clear, grouped insights.

Apply formula and string replacer in Knime to harmonize data sources, perform feature engineering, and compute metrics like body mass index.

Explore discretization, quantization, and binning in Knime by splitting numerical data into bins, converting to categorical data, and naming bins using auto binner or manual boundaries.

Learn the metanode concept and data generation in Knime, and explore splitting, merging, and collapsing workflows while generating synthetic data with various data generators.

Master splitting and merging data in Knime with metanodes, using column combiner and cell splitter to create or separate name and address fields from generated datasets.

Encode string labels into numbers via string replacer and string to number converter to enable numeric operations, demonstrating type conversion of sex from f/m to 1/0 and back.

Explore Bayes theorem and the naive Bayes model, including conditional probability and the assumption of feature independence, and apply them to a Knime practicum with a real dataset.

Explore Knime's naive bayes practicum by building a workflow from scratch, with binning and preprocessing, handling missing values, partitioning data, and evaluating with a confusion matrix and accuracy.

Explore decision tree algorithms, including information gain and the Gini index, and learn to implement them in Knime through a practical practicum.

Build and evaluate a decision tree workflow in Knime, including training/testing split, learner/predictor configuration, and pruning considerations for clearer outcomes.

Apply svm in Knime to classify gender from height and weight using a 66/34 train-test split, exploring linear, polynomial, cubic, and rbf kernels.

Build an end-to-end data science practicum in knime by training naive bayes, decision tree, knn, and svm on the iris dataset, with evaluation and extensions to apriori, clustering, and regression.

Explore logistic regression as a binary classification method powered by the sigmoid (logistic) function, mapping inputs to 0–1 probabilities and using a threshold for class predictions.

Practice end-to-end data science with knime by building a transaction dataset, applying apriori, and uncovering frequent item sets and association rules with minimum support.