Introduction

Where to find the ‘General Purpose Widgets’

The ‘General Purpose Widgets’ are all listed at the top of the ‘Add Widget’ drop down menu as shown in the below screenshot. The only exception to this is the ‘Chart Builder’ which can be enabled/disabled using the toggle button at the bottom of the ‘Widget’ sidebar as shown below.

The General Purpose Widgets…

Data Table

• Purpose: Displays the currently active dataset. 

• Widget Properties:

• Display Metadata Tables: When set to ‘Hide’ only a ‘Dataset’ table is shown with all user defined labels displayed (e.g. Judge names, Product names, Attribute names, Attribute category labels). When set to ‘Show’ an additional four tables are shown displaying the underlying EyeOpenR data structure including metadata tables.

• Output: The following four tables are displayed

• Dataset [Labelled]

• Dataset (Ref.): shows reference values (i.e. Judge Ref., Product Ref., Attribute Ref., Attribute category label Ref.) rather than user defined labels

• Attributes (Attribute Metadata)

• Assessors (Assessor/Panelist/Judge Metadata)

• Products (Product Metadata)

• Important Additional Information: If widget filters & groups are not active the dataset shown will be that resulting from any global filters or groups. If widget filters or groups are enabled then the data displayed will be the result of these widget level filters or groups.

• Example Display:

Project Metrics

• Purpose: Display simple metrics on your board (e.g. number of products or panelists). Project Properties (e.g. Products, Panelists, Attributes, Replicas, Sequences, Sessions) are clickable to reveal a list those that the widget currently has access to (i.e. after any global or widget level filters are applied).

• Widget Properties: 

• Choose Frequency Metric Boxes to display: This list displays ‘Attributes’ with ‘datatype’ equal to ‘binary’ or ‘category’. A sum of all values recorded for this attribute will be displayed in the corresponding metric box.

• Choose Means Metric Boxes to display: This list displays ‘Attributes’ with ‘datatype’ equal to ‘binary’ or ‘category’. A sum of all values recorded for the given attribute will be displayed in the corresponding metric box.

• Choose Project Property Metric Boxes to display: This list displays ‘Attributes’ with ‘datatype’ equal to ‘interval’, ‘integer’, ‘nominal’, ‘jar’, ‘scale’, ‘ordinal’, ‘liking’ or ‘category’. The mean of all values recorded for the given attribute will be displayed in the corresponding metric box.

• Choose Colour of Metric Boxes to display: Choose the colour pallet used for the metric boxes backgrounds.

• Number of Columns: The metric boxes you have selected will be split across the number of columns defined here when displayed.

• Output: The metric boxes you have defined in the widget properties (or default selection) will be displayed in a grid on the widget ‘Display’ panel.

• Important Additional Information: You can resize the Widget plus drag & drop the metric boxes to arrange them in the size and order you prefer.

• Example Display:

Text

• Purpose: Allows you to upload .txt or .docx files containing text that you can then display on your board.

• Widget Properties: 

• Title: The text entered here will be displayed in bold above any additional text body at the top of your text widget

• Font Size: This is the font size of the main body of text (i.e. text uploaded from docx or txt file)

• Text File: This allows you to choose a txt or docx file containing any text you wish to display. After the text file is uploaded, the text will be extracted and then displayed in the widget ‘Display’ once you click the ‘Update text in Widget’ button.

• Text editor: Edit/Paste the text you wish to add in this box before pressing ‘Update text in Widget’ to update the text in the ‘Display’.

• Output: 

• A text title and text body will be displayed defined by the above widget properties.

• Important Additional Information: Text formatting within the .txt or .docx document will largely be removed when uploading. However, the ability to preserve the formatting before it is displayed are planned for future versions of EO Insight. It is however currently possible for you to paste and format text within the text editor box within the widget properties.

• Example Display:

Image

• Purpose: Allows you to upload .png or .svg files containing images that you can then display on your board.

• Widget Properties: File upload button allows you to select the .png or .svg file containing the image you wish to display.

• Output: The uploaded image will be displayed and will auto adjust to fit the size of your widget should you resize the widget.

• Example Display:

Geographical Map

• Purpose: Displays region specific data on an interactive World Map (zoom or hover over regions on the map to reveal additional information). This widget allows you to create a choropleth, where regions/nations are coloured or shaded according to the values assigned to the given region/nation.

• Widget Properties: 

• Geographical Region Variable: Use this option to select the variable that contains the region to which values should be assigned. The variable defining the region/nation in the dataset must be coded using ISO-3 country codes. These region codes can either be stored as label values recorded in the Attributes sheet for a categorical attribute or in the form of a text value in each row of the ‘Data sheet’ for a ‘text’ attribute. See the Region variable in the attached demo data excel file as an example of how to code regional data for use with the Geographical Map widget.

• Variable to Plot: This is the variable that will be plotted on the map for each region. The value plotted for a region is defined by the ‘Measure to plot’ option below.

• Measure to Plot:

• Data Row Count: This simply counts the number of rows for this variable per region.

• Data Row Proportion: This records the proportion of rows in the dataset registered to a given region.

• Count Unique Values: This records the number of unique values of the ‘Variable to Plot’ for the given region.

• Mean: This displays for each region the arithmetic mean value of the ‘Variable to Plot’.

• Proportion: For each region the sum of the ‘Variable to Plot’ is calculated and then divided by the total sum across all regions of the ‘Variable to Plot’.

• Sum: For each region the sum of the ‘Variable to Plot’ is calculated and displayed.

• Colour Mode:

• Distinct: When selected a distinct colour (independent of the ‘Variable to Plot’) is assigned to each region

• Gradient: When selected regions are shaded according to their value (as defined by ‘Variable’ and ‘Measure’ to plot described above)

• Initial Longitude: A value defining the longitude on which the map is initially centered.

• Initial Latitude: A value defining the latitude on which the map is initially centered.

• Initial Zoom: A value defining initial zoom level of the map (10 = maximum zoom, i.e. closest to the ground).

• Output: 

• Map: The interactive map displaying region specific values

• Table: A table containing a row for each region for which there is data and the corresponding value assigned to the given region.

• Important Additional Information: To export the map you can click the ‘camera’ icon displayed on the right side of the map. The image of the map generated will be centered & zoomed according to the properties defined in the Widget Properties described above (i.e. initial longitude, initial latitude & initial zoom).

• Example Display:

Pie Chart

• Purpose: Creates Pie / Donut charts of variables of datatype “nominal”, “jar”, “scale”, “ordinal”, “binary” or “liking” to display on your board. 

• Widget Properties: 

• Attribute: Select from a list of suitable attributes which attribute’s results will be displayed in the pie charts

• Split Parameter: Choose whether you wish to display a single pie chart summarising the entire dataset (i.e. ‘No Split’ option), or you wish to have a separate pie chart for each Product, Session, Replica or Sequence.

• Choose Labelling: decide whether pie charts should display for each category the category label and/or the value or percentage

• Output: 

• Pie Chart: Image(s) of your chart(s) with labels, title and legend

• Table: A table showing the values for each category (if a split is applied, a separate row will be shown for each split)

• Important Additional Information: If the charts or labels overlap, resize the widget in order for the plot to adjust to the new dimensions. 

• Example Display:

Distribution

• Purpose: Generates a density plot for an attribute of type ‘integer’ or ‘interval’ to display on the board.

• Widget Properties: 

• Attribute: Select the ‘integer’ or ‘interval’ attribute for which you wish to display distributions

• Split Parameter: Choose whether you wish to display a single distribution summarising the entire dataset (i.e. ‘No Split’ option), or you wish to have a separate distribution plotted for each Product, Session, Replica or Sequence.

• Output: 

• Distribution Plot: Distribution plots of the selected ‘Attribute’

• Table: Table displaying the raw data fed to the geom_density() function.

• Important Additional Information: Distributions are generated with the geom_density() function from the ggplot2 R package with parameter adjust = 1.5. 

• Example Display:

Histogram

• Purpose: Generates a histogram for an attribute of type ‘integer’ or ‘interval’ to display on the board.

• Widget Properties:

• Attribute: Select the ‘integer’ or ‘interval’ attribute for which you wish to create histograms

• Split Parameter: Choose whether you wish to display a single histogram summarising the entire dataset (i.e. ‘No Split’ option), or you wish to have a histogram plotted split by Product, Session, Replica or Sequence.

• Output:

• Histogram: Histogram plot of the selected ‘Attribute’

• Table: Table displaying the raw data fed to the geom_histogram() function.

• Important Additional Information: A count records each row in the dataset for which the value is recorded. Histograms are generated with the geom_histogram() function from the ggplot2 R package

• Example Display:

Violin Plot

• Purpose: Generate violin plots to quickly understand the distribution of your data. Plot a single attribute (of datatype ‘’interval, ‘jar’, ‘scale’, ‘ordinal’, ‘integer’ or ‘liking’) split by Product, Judge, Session, Replica or Sequence or plot all attributes. 

• Widget Properties: 

• Attribute: Select the attribute for which you wish the violin plots will be generated (if split parameter is set to ‘Attributes’ or ‘Attributes [Standardised]’ then this parameter is ignored and all attributes are plotted)

• Split Parameter: Choose whether you wish to plot a single attribute split by Product, Judge, Session, Replica or Sequence or plot all attributes (i.e. ‘Attributes’ or ‘Attributes [Standardised]) across the entire dataset (incl. widget or global filters & groups). If ‘Attributes Standardised’ is selected then variables will be standardised before generating the violin plots.

• Output: 

• Violin Plot(s): Violin plots generated using plot_ly() with type = ‘violin’.

• Table: Table displaying the raw data fed to the plot_ly() function

• Important Additional Information: Violin plots generated using R package plotly.

• Example Display:

Design Summary

• Purpose: Quickly generate tables to inspect the experimental design and the structure of the collected data. These tables display the number of observations per cell across combinations of factors (e.g. Product, Assessor, Session, Sequence, Replica). Colour coding highlights imbalances or missing data, helping users easily identify irregularities across Panelists/Assessors/Judges, Products, Sessions, Sequences, and Replicates.

• Widget Properties: There are no widget specific parameters for the Design Summary Widget.

• Output: 

• By Product | Judge: Each cell shows how many evaluations a judge made for that product; the row Total is the product’s total evaluations.

• By Product | Session: Each cell shows how many times a product was evaluated in that session; the row Total is the product’s total across sessions.

• By Product | Sequence: Each cell shows counts of a product appearing at a given position/order; the row Total is the product’s total appearances.

• By Product | Replica: Each cell shows how many observations for a product for each replicate; the row Total is the product’s total across replicates.

• By Assessor | Session: Each cell shows how many samples that assessor evaluated in that session; the row Total is that assessor’s session workload.

• By Assessor | Sequence: Each cell shows how many evaluations an assessor made at each order position; the row Total reflects that assessor’s overall counts.

• By Assessor | Replica: Each cell shows how many evaluations an assessor made for each replicate; the row Total is that assessor’s total across replicates.

• Important Additional Information: Cells are shaded from green (low number of observations) to white (medium number of observations) to red (high number of observations), reflecting the relative number of observations in each cell.

• Example Display:

Data Summary

• Purpose: Quickly visualise the structure of all variables in your dataset. This widget generates a visualisation of the structure of each variable in your dataset plus summary statistics (e.g. proportion of missing data, mean, median, sd).

• Widget Properties: There are no widget specific parameters for the Data Summary Table Widget.

• Output: 

• Data Summary: A table consisting of a row per variable in the dataset. For categorical variables all unique values are listed, number of categories defined and a visual representation of the distribution of categories across the dataset. For numeric variables histograms are displayed showing the spread across the variables range, in addition the mean, median and sd are provided. The proportion of missing values is also provided for all variables.

• Important Additional Information: The data summary widget executes the gt_plt_summary() function from the gtExtras R package on the dataset accessible to the widget (incl. Global or Widget level filters & groups).

• Example Display:

NaileR

• Purpose: 

• This method provides a quick, broad, preliminary analysis of your dataset. It aims to understand and describe how, for example, different product types (i.e. Factor A) and possible test conditions or panels (i.e. Factor B) influence responses (i.e. Endogenous Variables), identifying where possible what is distinctive about each product (i.e. Factor A).

• Method:

• The widget achieves this by implementing the nail_qda function from the NaileR package. It first performs a QDA analysis using the 'decat' function from the SensoMineR package with the following parameters (adjustable via Widget's Properties tab):

• Formula = 'Factor A' * 'Factor B'

• Significance Level = 0.05

• Endogenous Variables = any variables of type: jar, scale, ordinal, liking, interval, integer

• The results from the QDA analysis are then passed, in addition to your 'Dataset Details' and 'LLM Task', to the AI Model (LLM provider: OpenAI) for evaluation & interpretation.

• Widget Properties: 

• Enable Switch: Enable the analysis of your data by the NaileR widget & OpenAI LLM. By enabling NaileR you agree to your data & prompts being shared with OpenAI! Your data will not be used to train or improve models.

• Factor A: A list of any categorical variables in your dataset. 

• Factor B: A list of any categorical variables in your dataset. 

• Dataset Details: This is your opportunity to describe the dataset and study. This information will be used by the LLM to tailor it’s response.

• AI Tool’s Task: This allows you to guide the LLM towards generating an output that achieves your intended goal. Describe what task you wish the AI tool to do given its access to your data and the output from the preliminary analysis.

• Significance Level: Defines the level at which a difference should be classified as significant.

• Output: Displays results generated by the LLM following the process outlined above. AI generated output may contain inaccuracies.

• Important Additional Information: 

• When using this function please be aware that your data & prompts will shared with OpenAI, also AI generated output may contain inaccuracies. 

• For further information on the R packages used in this widget see:

• NaileR: Interpreting Latent Variables with AI, Nel Hervé & Sébastien Lê

• SensoMineR: A package for sensory data analysis, S. Le & F. Husson (2008)

• Example Display:

CATA Bar Chart

• Purpose: Quickly generate bar charts for both product related & non-product related CATA data. 

• Widget Properties: 

• Attribute: Choose the name of the ‘Parent Question’ of the CATA data (‘Parent Questions’ are defined by the ‘Parent_Question’ column on the dataset’s ‘Attributes’ metadata sheet).

• Split Parameter: Choose whether or not (i.e. No Split) to split the results by Product, Session, Replica or Sequence.

• Measure: Choose either the ‘Count Total’ which provides the total number of occasions on which a given option was selected or ‘Proportion’ which records the total number of occasions on which a given option was selected divided by the total number of occasions in which panelists could choose the given option.

• Output: 

• Chart: Bar chart of CATA question data

• Table: Table recording the data plotted in the Bar Chart

• Important Additional Information: As stated above the ‘Attribute’ available to plot is defined by the presence of data in the ‘Parent_Question’ column on the dataset’s ‘Attributes’ metadata sheet (as shown in the screenshot below). All options for the same CATA question must share the same value in the ‘Parent Question’ column. Should you wish to assign response options to a parent attribute import a dataset with appropriate values in the ‘Parent_Question’ column (to add to your current dataset first export the raw dataset to excel [i.e. Export -> Download Current Dataset], edit the excel and then import the edited excel file [i.e. Data Sources -> Load External Dataset]). It is also possible to define/adjust the ‘Parent_Question’ on the ‘Attributes' tab within the ‘Advanced Import Wizard’ when importing an external dataset (see Import Data Knowledge Base Article).

• Example Display (CATA Bar Chat widget and additional ‘Data Table’ widget):

Chart Builder

• Purpose: Build a wide range of different charts from your dataset to either export for use elsewhere or to import to an ‘Image’ widget on your EO Insight board.

• How to Enable Chart Builder:

• You can enable the ‘Chart Builder’ by clicking the switch in the ‘Chart Builder’ section of the ‘Widget’ sidebar to ‘On’ as shown below. Once enabled the Chart Builder will always be placed at the top of your board. It can be removed by toggling the ‘Enable Chart Builder’ switch to ‘Off’.

• How to Build your Chart: 

1. Drag and drop the variables from your dataset (Red Box) into the grey dotted boxes (Orange Box) above the chart panel, to construct your plot which will display in the chart panel in the center of the Chart Builder (Blue Box):

2. Choose the chart type by clicking the drop down in the top left corner (Green Box):

3. Add additional layers to your chart using the different ‘Geom’ tabs available at the top of the Chart Builder (Purple Box):

4. Filter your data or adjust the size, colour scheme, labels, titles, and many other properties of your chart using the menus below the chart (Yellow Box):

• How to Export your Chart: Export your chart by clicking the ‘Download’ button in the top right corner of your chart panel (Pink Box), then select the format: 

• How to Use your Chart in EO Insight: After exporting your chart as a png or svg (recommended) image you can import it to any ‘Image’ widget you have added to your board. 

• Important Additional Information: The ‘Chat Builder’ in EO Insight is created using the ‘esquisse’ R package.

• Example Display: