Icon

02_​Guided_​Exploration

Guided Exploration

Guided exploration provides a data scientist with insights into data at hand. Any dataset can be used. The data are automatically cleaned by removing columns with constant values and by identifying the correct data types of the columns. The relationships between and within columns are calculated and shown by charts in an interactive composite view, nwhere columns can also be flagged. By default the flagged columns are excluded from the dataset. If any other action for the flagged columns is preferred, a new action can be defined by replacing the Reference Column Filter node.

Guided ExplorationThis workflow defines a fully automated web based application for the KNIME WebPortal to show relevant visualizationiteration after iteration. The workflow was designed for data scientist to easily create a dashboard and find relationshipsbetween columns. The wrapped metanode "Guided Exploration" is nested in a recursive loop and outputs a web page ateach iteration. If the data scientist discards columns the dashboard is updated in the next iteration. Only the first twoiterations can be executed from KNIME Analytics Platform using the command "Do one loop step" (Ctrl + Alt + F6). The Process Step by Step1. Upload your data / Select one of the available datasets2. Inspect the visualizations in "Guided Exploration" Wrapped Metanode3. Select at the bottom of the View the columns you would like to remove4. Apply the settings and Close the View 5. "Do one loop step" (Ctrl + Alt + F6) on the Recursive Loop End6. Reopen the Wrapped Metanode view Guided Exploration Dashboard Data Source and Licenseadult.csv : archive.ics.uci.edu/ml/datasets/adultflights.csv : stat-computing.org/dataexpo/2009/the-data.htmlsales.csv : kaggle.com/kyanyoga/sample-sales-data remove columns ReferenceColumn Filter RecursiveLoop Start Recursive Loop End Variable toTable Row Inject Variables(Data) CASE SwitchData (End) sanity check sanity check end cycle Pre-Processing Guided Exploration Upload Guided ExplorationThis workflow defines a fully automated web based application for the KNIME WebPortal to show relevant visualizationiteration after iteration. The workflow was designed for data scientist to easily create a dashboard and find relationshipsbetween columns. The wrapped metanode "Guided Exploration" is nested in a recursive loop and outputs a web page ateach iteration. If the data scientist discards columns the dashboard is updated in the next iteration. Only the first twoiterations can be executed from KNIME Analytics Platform using the command "Do one loop step" (Ctrl + Alt + F6). The Process Step by Step1. Upload your data / Select one of the available datasets2. Inspect the visualizations in "Guided Exploration" Wrapped Metanode3. Select at the bottom of the View the columns you would like to remove4. Apply the settings and Close the View 5. "Do one loop step" (Ctrl + Alt + F6) on the Recursive Loop End6. Reopen the Wrapped Metanode view Guided Exploration Dashboard Data Source and Licenseadult.csv : archive.ics.uci.edu/ml/datasets/adultflights.csv : stat-computing.org/dataexpo/2009/the-data.htmlsales.csv : kaggle.com/kyanyoga/sample-sales-data remove columns ReferenceColumn Filter RecursiveLoop Start Recursive Loop End Variable toTable Row Inject Variables(Data) CASE SwitchData (End) sanity check sanity check end cycle Pre-Processing Guided Exploration Upload

Nodes

Extensions

Links