Individual product dashboard¶

Work in progress

This site is still under construction. Please come back later for further information and documentation.

The product dashboards are available as soon as a product has been evaluated. They will also automatically update to include the most recent results.

Accessing a dashboard¶

To access a product dashboard, go to Dashboards in the left side bar or from the top right menu . Navigate to the PRODUCT DASHBOARDS section, and click View results in the table next to the product you are interested in. You can also directly search for a specific product in the search bar at the top of the table.

Dashboards should be displayed on a big enough screen

Although dashboards can be accessed from a mobile device such as a smartphone, there must be enough screen space needs to view the charts. We strongly recommend that you view the dashboards on a desktop or tablet screen, as a smartphone tends to cut off the charts, making them difficult to read and interact with.

Overview¶

Product dashboards aims at displaying qualities and flaws of a given product. They can also be used to visualize the evolution of its sensory profile with time (for example during maturation).

The data can be filtered based on tasting dates and assessors. In the Apply filters section at the top of the dashboards, you can include/exclude assessors from the results, as well as select only a given period. Click on Apply to update the charts.

Different tasting grids

It is possible that over time, different tasting grids will be used to evaluate the same sample. In that case, the results of each tasting grid will be displayed separately. The name of the corresponding tasting grid is shown above the charts.

TL;DR

The first chart on the page includes all the tasting grid found in the results if there were multiple.
- Scores per criterion (if the product was only tasted once) shows the scores per criteria given by each single taster;
- Evolution of average score and range (if the product was tasted more than once over time) shows the average total score of the product, as well as the range of scores among the assessors (lowest and highest total score);
The following charts are related to a single tasting grid; the name of the tasting grid will be displayed above the charts.
- Evolution per criterion displays the average score for each criterion of the tasting grid. If the product was tasted multiple times, the score progression between the oldest and most recent tastings will be specified;
- Scores per criterion shows the averaged score for each criterion of the grid on a so called radar or spider chart for easier interpretation;
- Tasting notes treemap shows the frequency of words found in the tasting notes corpus.
- Principal Components Analysis with components as linear combinations of tasters displays the loading plots of a Principal Components Analysis on the total average scores given by each taster, and aims at showing agreements and disagreements between tasters;
- Principal Components Analysis with components as linear combinations of tasting criteria displays the loading plots of a Principal Components Analysis on the total average scores for each criteria, and aims at showing patterns and relations between sensory characteristics;

The charts are interactive

The charts can be interacted with in a number of ways. The interactions depend on the chart type but in general you can:

Hover over the chart to see more details about the numbers;
Click and Drag to zoom in on a specific area of the graph; scroll down to zoom out;
Clicking on the legend activates/deactivates that part of the graph.

Charts details¶

Scores per criterion¶

The scores per criterion charts is displayed only if the product has been evaluated on one single instance (independently of the number of assessors), i.e. if the database contains results only for one single date.

It displays the scores given by each assessor for every criteria as columns, as well as the corresponding average score as line. Note that the scores are normalized to 10 for ease of comparison: even if the available minimum and maximum scores of a given criterion were different on the tasting grid, they are all displayed as from 0 to 10 on the chart.

What to look for

This is a first quick glance at the strength and weaknesses of the product. It also already gives an idea on the sensitivities of assessors and how they agree or disagree.

Evolution of average score and range¶

This chart shows the evolution of the average, maximum and minimum scores given over time by assessor. In case different tasting grids have been used, each line represents a given protocol. The solid colour line is the average, while the lighter colour area represents the maximum and minimum scores given at each specific date.

Note that the scores are expressed as a percentage of the maximum possible score of the considered tasting grid. As multiple tasting grids can have been used over time, and considering that two tasting grids can have completely different scales (e.g. one grid allowing scores from 0 to 50 while the other allows scores from 0 to 120), to allow for a meaningful comparison, they are all normalized to the same scale of 0 to 100.

What to look for

Considering that a sensory analysis is a measure of liking and disliking, the evolution chart can help determine the optimum sensory profile of a given product, or reveal when a change in the product brings improvement or deterioration of the profile. The area around the line can also help identify whenever a group of assessor concur (translated as a narrow area around the line), or on the contrary when the product divides the assessor (translated as a wide area around the line).

Evolution per criterion¶

The aim of the evolution per criterion is to provide details on how each sensory criterion is evolving over time; it gives the total average score as well as the average score per criterion on the most recent date considered, and the evolution since the oldest date considered.

The scores are displayed without normalization, i.e. as per the tasting criterion scale. Note that the displayed values depend on the selected period in the date range filter.

What to look for

Using the global evolution of average score and range above, the chart can provide more details and explain which criteria are responsible for any increase or decrease in the total score. Use the date range filter around a spike on the previous chart to highlight and understand precisely how the sensory profile is evolving.

Scores per criterion and evolution over the selected period¶

Similarly to the previous graph, the radar chart shows the average score per criteria of the tasting grid and its evolution at a glance; the blue web gives the scores at the oldest date, while the red web gives the scores at the most recent tasting date. Note that the scores here are normalized on a scale from 1 to 10 for easier comparison between criteria: even if a sensory characteristic is evaluated on a different scale on the tasting grid, its score here will be shown between 0 and 10. As such the radar chart does not take into account the weight given to each criteria on the tasting grid.

What to look for

You can quickly see which were the criteria which have displayed the highest changes over the considered period, as well as find out what are the qualities and flaws of a product.

Wordcloud¶

The word cloud graph is a treemap of the words found in the corpus of tasting notes (all tasters included); it shows the relative frequency or importance of the different vocables within the corpus.

Depending on the tasting protocol, different colours refer either to different categories of criteria (e.g. Nose, Palate, General, etc.) or to different criteria (e.g. Expressiveness in the nose, Flavour complexity on palate, etc.). A single colour indicates general tasting notes which are not related to a specific criterion.

Removed words

The text corpus is first trimmed of some of the most common words (such as "the", "and", "this", etc) that are not expected to provide any meaningful information.

Date range

The wordcloud takes into account all the tasting notes over the selected period. If the sensory profile has evolved a lot over the considered period, it might be a good idea to select a single date in the date range filter to analyse the words, as the comments have most probably changed as well.

Principal component analysis¶

The principal component analysis (PCA) aims at reducing the number of variable in the dataset in order to find patterns and relationships in the data. Applied to sensory evaluation, two sets of variables are considered: on one side, we are looking at the sensory criteria as variables, considering then only the consensus scores (i.e. the average scores among all the assessors), and on the other side, we are looking at the assessors themselves as variables, considering the total score they are each giving to a product.

The goal is then to highlight agreements between assessors or group of assessors, and to find out how sensory criteria are linked to each others and meaningful in the sensory evaluation as a whole.

More details can be found in the PCA section of Reading the charts.

Minimum amount of data required

Performing a PCA is only available if enough data has been collected. The charts will remain empty in such a case, displaying a message that not enough data is available. Keep in mind however that depending on the case considered, PCA can be calculated on either of the tasting criteria or assessors, both, or none.

A sample that has been evaluated only once by multiple tasters will give information on their agreement but not on the relevance of the tasting criteria. A sample evaluated multiple times by one single taster will of course give information only on the relevance of the tasting criteria.

General considerations¶

Whether the PCA is performed on the tasters or on the sensory criteria, it is capital to pay attention to the explained variance, represented by the area chart at the top.

As the bottom chart plots the value of Principal Component 1 (PC1) on the x-axis and Principal Component 2 (PC2) on the y-axis, the explained variance provides information on the relative importance of both the components in explaining the global variation among the data. The explained variance chart plots the cumulative explained variance of the components. Note that said importance is always decreasing, i.e. the explained variance of PC1 is always greater than the explained variance of PC2, which is always greater than the explained variance of PC3, and so on. In other words, the greater the value at PC2, the more reliable is the chart under it. For the same reason, the difference between the explained variance of PC1 and PC2 provides information on the relative importance of one to the other.

For example, in the illustrative graph above, PC1 and PC2 together explains 97% of the variation within the dataset, which is a very good result. The corresponding plots will be highly informative in finding patterns and relations between the variables. However, PC1 on its own already explains 93% of the variation. The differences along the x-axis (horizontally) will then be much more important in understanding the variation than the differences along the y-axis (vertically). If PC1 was only explaining 50% of the variation, with PC2 explaining the remaining 47%, both should be looked at with equal care.

PCA on tasters¶

The first PCA performed provides information on the agreement between tasters and indications on how distinct the sample has seemed to each of them every time it has been evaluated.

What to look for

There are 4 cases to consider:

Tasters in good agreement will be close to each others. You therefore might be able to find patterns between group of tasters who seem to share the same opinion;
Tasters with strong disagreement will lie on opposite side of a straight line through the origin of the chart. It would reveal that their overall opinion on the sample over time has been quite diverging, liking and disliking being reversed;
Tasters who do not specially agree nor disagree will lie relatively far from each other, but not on opposite side of a straight line as above;
Finally, tasters who give very similar opinions from one evaluation to the other and whose opinion do not really impact the variation between evaluations will fall close to the center. It could be due to a poor use of the scales in the tasting protocols, or because the sample shows very little variation to them.

PCA on tasting criteria¶

The second PCA aims at finding links between the evaluated sensory characteristics and how relevant they are in differentiating between evaluations.

PCA on criteria — PCA performed on sensory criteria as features. Each of the points are representative of one of the criteria of the tasting grid

What to look for

As for the PCA on tasters, there are 4 cases to consider:

Criteria strongly positively correlated (i.e. which usually improve or recede concurrently) will be close to each other;
Criteria strongly negatively correlated (i.e. an improvement in one will cause a decline in the other) will lie on opposite side of a straight line going through the origin of the chart.
Criteria which do not show relation to each other will be relatively far to each other, but not opposed as above.
Criteria which do not vary a lot between evaluations will lie close to the center (they might be discarded for further analysis)