🌌🏭KQL Reloaded: First Steps (Part VII: Basic Data Visualizations)

One of the greatest aspects of KQL and its environment is that creating a chart is just one instruction away from the dataset generated in the process. Of course, the data still need to be in an appropriate form to be used as source for a visual, though the effort is minimal. Let's consider the example used in the previous post based ln the ContosoSales data, where the visualization part is everything that comes after "| render":

// visualizations by Country: various charts
NewSales
| where SalesAmount <> 0 and ProductCategoryName == 'TV and Video'
| where DateKey >=date(2023-02-01) and DateKey < datetime(2023-03-01)
| summarize count_customers = count_distinct(CustomerKey) by RegionCountryName
| order by count_customers desc
//| render table
//| render linechart
//| render areachart 

//| render stackedchart
//| render columnchart
| render piechart
    with (xtitle="Country", ytitle="# Customers",
    title="# Customers by Country (pie chart)", legend=hidden)

Output:

# Customers by Country (various charts)

It's enough to use "render" with the chart type without specifying the additional information provided under "with", though the legend can facilitate data's understanding. Unfortunately, the available properties are relatively limited, at least for now. 

Adding one more dimension is quite simple, even if the display may be sometimes confusing as there's no clear delimitation between the entities represented while the legend grows linearly with the number of points. It might be a good idea to use additional charts for the further dimensions in scope. 

// visualizations by Region & Country: various charts
NewSales
| where SalesAmount <> 0 and ProductCategoryName == 'TV and Video'
| where DateKey >=date(2023-02-01) and DateKey < datetime(2023-03-01)
| summarize count_customers = count_distinct(CustomerKey) by ContinentName, RegionCountryName
| order by count_customers desc   
//| render stackedareachart 
//| render linechart 
//| render table 
//| render areachart 
//| render piechart
| render columnchart 
    with (xtitle="Region/Country", ytitle="# Customers",
    title="#Customers by Continent & Country", legend=hidden)

Output:

# Customers by Continent & Country (column chart)

Sometimes, it makes sense to reduce the number of values, recommendation that applies mainly to pie charts:

// visualizations by Zone: pie chart
NewSales
| where SalesAmount <> 0 and ProductCategoryName == 'TV and Video'
| where DateKey >=date(2023-02-01) and DateKey < datetime(2023-03-01)
| summarize count_customers = count_distinct(CustomerKey) by iif(RegionCountryName in ('United States', 'Canada'), RegionCountryName, 'Others')
| render piechart
    with (xtitle="Country", ytitle="Sales volume",
    title="Sales volume by Zone")

Output:

# Customers by Zone (pie chart)

Adding a second set of values (e.g. Total cost) allows to easily create a scatter chart:

// visualization by Occupation: scatter chart
NewSales
| where SalesAmount <> 0 and ProductCategoryName == 'TV and Video'
| where DateKey >=date(2023-02-01) and DateKey < datetime(2023-03-01)
| summarize count_customers = count_distinct(CustomerKey) 
    , TotalCost = sum(TotalCost) by Occupation
| order by count_customers desc
| render scatterchart 
    with (xtitle="# Customers", ytitle="Sales volume",
    title="# Customers vs Sales volume by Occupation", legend=visible )

Output:

# Customers vs Sales volume by Occupation (scatter chart)

The visualizations are pretty simple to build, though one shouldn't expect that one can build a visualization on top of any dataset, at least not without further formatting and eventually code changes. For example, considering the query from the previous post, with a small change one can use the data with a column chart, though this approach might have some limitation (e.g. it doesn't work pie charts):

// calculating percentages from totals: column chart
NewSales
| where SalesAmount <> 0 and ProductCategoryName == 'TV and Video'
//| where DateKey >=date(2023-02-01) and DateKey < datetime(2023-03-01)
| summarize count_customers = count_distinct(CustomerKey)
    , count_customers_US = count_distinctif(CustomerKey, RegionCountryName == 'United States')
    , count_customers_CA = count_distinctif(CustomerKey, RegionCountryName == 'Canada')
    , count_customers_other = count_distinctif(CustomerKey, not(RegionCountryName in ('United States', 'Canada')))
| project Charting = "Country"
    , US = count_customers_US
    , CA = count_customers_CA
    , other = count_customers_other
| render columnchart

    with (xtitle="Region", ytitle="# Customers",
    title="# Customers by Region")

Output:

# Customers by Region (column chart)

There are a few more visuals that will be considered in a next post. Despite the relatively limited set of visuals and properties, the visualizations are useful to get a sense of data's shape, and this with a minimum of changes. Ad-hoc visualizations can help also in data modeling, validating the logic and/or identifying issues in the data when creating the queries, which makes it a great feature. 

Happy coding!

Previous Post <<||>> Next Post

🌌🏭KQL Reloaded: First Steps (Part I: Simple Queries)

If one has followed the Microsoft training, webcasts and other resources, it becomes clear that the Kusto Query Language (KQL) can't be ignored as it's useful in various scenarios thar deal with large volumes of data. 

Even if KQL was created to query big tables, at least for the first steps it's recommended to start with a small data model, and when the basics were understood one can move to bigger databases. Moreover, it's easier to validate the logic when using small datasets.

Probably, the most important aspect before starting is that KQL is case-sensitive and this applies to everything – table and column names, operators, functions, etc. The below code can be tried in the online explorer (see [2], [3]), which makes available several databases for playing around. The users might need to register beforehand.

The following queries are based on the ContosoSales database (available in the above mentioned link). First, here are some simple projections. Each query is preceded by its short description in which the text was commented via "//" and must be run individually. 

// selecting all records
Customers 

// selecting all records
Customers 
| take 10

// multiple filters
Customers 
| where CityName == 'Berkeley'
| where Occupation != 'Professional'
| take 10

// multiple filters on the same column
Customers 
| where Occupation == 'Professional' and Occupation == 'Clerical')
| take 10

// multiple filters on the same column
Customers 
| where Occupation in ('Professional','Clerical')
| take 10

// multiple filters on the same column
Customers 
| where not(Occupation in ('Professional','Clerical'))
| take 10

//subset of columns
Customers
| take 5
| project ContinentName, CityName, FirstName, LastName, Gender

Here are some example for the selection of unique values, the equivalent of SELECT DISTINCT from SQL:

//distinct values used
Customers
| distinct  Occupation

//distinct values used sorted ascendingly 
Customers
| distinct  Occupation
| sort by Occupation asc

//combinations of values used
Customers
| distinct  Occupation, Education
| sort by Occupation asc, Education asc

When further data is needed, one needs to resume to grouping values, the equivalent of GROUP BY:

// record count
Customers 
| count

// record count for constraint
Customers 
| where CityName == 'Berkeley'
| count

// record count for constraint: returns 0 records (KQL is case sensitive)
Customers 
| where CityName == 'BERKELEY'
| count

// numnber of records by occupation 
Customers
| summarize occupations_count = count() by Occupation

// numnber of records by occupation with bar chart visual
Customers
| summarize occupations_count = count() by Occupation
| render barchart

The last query renders the data directly to a bar chart, which is a cool feature, especially when is needed to understand the distribution of values. Executing the query without the last line renders the initial dataset.

Azure Data Explorer - Chart Example

Here are some first impressions:
1) The language is relatively simple, though the transition from SQL to KQL requires time, even if the thinking process of writing the code is not that far away. For those with no SQL knowledge, the transition might be a bit more challenging, though practice makes perfect!
2) One can try to run the code line by line to understand the steps used.
3) There are also some online code converters from SQL to KQL (see for example msticpy).
4) The intellisense capabilities facilitate the overall experience. 
5) Unfortunately, there seems to be no code formatter to HTML for KQL, so one needs to compromise in one way or another.
6) For a comparison between SQL and KQL see [4].

Happy coding!

|>> Next Post 

References:
[1] Microsoft Learn (2024) Kusto Query Language learning resources [link]
[2] TestingSpot Blog (2021) Learning Kusto Query Language - A tool for performance test engineers [link]
[3] Azure Data Explorer (2025] Create a free cluster to explore your data [link]
[4] Microsoft Learn (2024) SQL to Kusto Query Language cheat sheet [link]

Resources:
[R1] GitHub repository (2022) Sentinel Queries [link]
[R2] GitHub repository (2022) Kusto Query Language Grammar [link]
[R3] GitHub repository (2024) The Definitive Guide to KQL source code [link]
[R4[ Git Hub repository (2022) Learning Kijo [link]
[R5] Kusto gamification [link]

🧭💹Business Intelligence: Perspectives (Part XIX: Data Visualization between Art, Pragmatism and Kitsch)

Business Intelligence Series

The data visualizations (aka dataviz) presented in the media, especially the ones coming from graphical artists, have the power to help us develop what is called graphical intelligence, graphical culture, graphical sense, etc., though without a tutor-like experience the process is suboptimal because it depends on our ability of identifying what is important and which are the steps needed for decoding and interpreting such work, respectively for integrating their messages in our overall understanding about the world.

When such skillset is lacking, without explicit annotations or other form of support, the reader might misinterpret or fail to observe important visual cues even for simple visualizations, with all the implications deriving from this – a false understanding, and further aspects deriving from it, this being probably the most important aspect to consider. Unfortunately, even the most elaborate work can fail if the reader doesn’t have a basic understanding of all that’s implied in the process.

The books of Willard Brinton, Ana Rogers, Jacques Bertin, William Cleveland, Leland Wilkinson, Stephen Few, Albert Cairo, Soctt Berinato and many others can help the readers build a general understanding of the dataviz process and how data visualizations or simple graphics can be used/misused effectively, though each reader must follow his/her own journey. It’s also true that the basics can be easily learned, though the deeper one dives, the more interesting and nontrivial the journey becomes. Fortunately, the average reader can stick to the basics and many visualizations are simple enough to be understood.

To grasp the full extent of the implications, one can make comparisons with the domain of poetry where the author uses basic constructs like metaphor, comparisons, rhythm and epithets to create, communicate and imprint in reader’s mind old and new meanings, images and feelings altogether. Artistic data visualizations tend to offer similar charge as poetry does, even if the impact might not appeal so much to our artistic sensibility. Though dataviz from this perspective is or at least resembles an art form.

Many people can write verses, though only a fraction can write good meaningful poetry, from which a smaller fraction get poems, respectively even fewer get books published. Conversely, not everything can be expressed in verses unless one finds good metaphors and other aspects that can be leveraged in the process. Same can be said about good dataviz.

One can argue that in dataviz the author can explore and learn especially by failing fast (seeing what works and what doesn’t). One can also innovate, though the creator has probably a limited set of tools and rules for communication. Enabling readers to see the obvious or the hidden in complex visualizations or contexts requires skill and some kind of mastery of the visual form.

Therefore, dataviz must be more pragmatic and show the facts. In art one has the freedom to distort or move things around to create new meanings, while in dataviz it’s important for the meaning to be rooted in 'truth', at least by definition. The more the creator of a dataviz innovates, the higher the chances of being misunderstood. Moreover, readers need to be educated in interpreting the new meanings and get used to their continuous use.

Kitsch is a term applied to art and design that is perceived as naïve imitation to the degree that it becomes a waste of resources even if somebody pays the tag price. There’s a trend in dataviz to add elements to visualizations that don’t bring any intrinsic value – images, colors and other elements can be misused to the degree that the result resembles kitsch, and the overall value of the visualization is diminished considerably.

Previous Post <<||>>  Next Post

📊Graphical Representation: Graphics We Live By (Part X: Pie and Donut Charts in Power BI and Excel)

Pie charts are loved and hated by many altogether, and there are many entitled reasons to use them and avoid them, though the most important criteria to evaluate them is whether they do the intended job in an acceptable manner, especially when compared to other representational means. The most important aspect they depict is the part to whole ratio, which even if can be depicted by other graphical tools, few tools are efficient in representing it. 

The pie chart works well as a visualization tool when it has only 3-5 values that are easily recognizable in the visualization, however as soon the size or the number of pieces vary considerably, the more difficult it is to visualize and interpret them, in case their representation has more negative than positive effects. There are many topics that form something like a long tail - the portion of the distribution having many occurrences far from the head or beginning. Displaying the items from the long tail together with the other components together can totally obscure the distribution of the items from the long tail as they become unrecognizable in the diagram. 

One approach to handle this is to group all the items from the long tail together under a piece (e.g. Other) and use a second form of representation to display them separately. For example,  Microsoft Excel offers a way to zoom in the section of a pie chart with small percentages by displaying them in a second pie chart (pie of pie) or bar chart (bar of pie), something like a "zoom in" perspective (see image below). Unfortunately, the feature seems to limit itself only to small percentages, and thus can't be used currently to offer a broader perspective. Ideally, it would be useful to zoom in on any piece of the pie, especially when the items are categorized as a hierarchy with two or even more levels. 

Unfortunately, even modern visualization tools offer limited features in displaying this kind of perspective into a flexible unitary visualization, and thus users are forced to use their creativity in providing proper solutions. In the below example the "Renewables" piece of pie is further broken down into several components of a full pie, an ensemble supposed to function as a single form of representation. With a bit of effort, the reader probably will understand the meaning behind the two pie charts, however the encoding of colors and other elements used are suboptimal in the decoding process. 

Pie Charts - Original Solution

In the above example, the arrow may suggest that in between the two donut charts exists a relationship, reflected also in the description provided, however the readers may still have difficulties in correctly interpreting the diagrams, especially when there's some kind of overlapping or other type of implied or unimplied resemblance. If the colors overlap or have other similarities, are they intentional? If the circles have the same size, does this observed resemblance have a meaning? The reader shouldn't bother himself with this type of questions, but see the resemblance and the meaning of the various elements with a minimum of effort while decoding a chart's elements. Of course, when the meaning is not clear, some guidance should be ideally provided!

Unfortunately, Power BI doesn't seem to have a similar visual like the one from Excel yet, however with a bit of effort one can obtain similar results, even if there are other minor or important limitations. For example, the lines between the two pie charts can't be drawn, so one is forced to use other encodings to show that there's a connection between the Renewable slice and the small pie chart. Moreover, the ensemble thus created isn't treated unitary and handled accordingly. Frankly, the maturity of a graphical representation environment can and should be judged also from this perspective!

The below representation built in Power BI uses a few tricks to display two pie charts together. The smaller pie chart representing the breakdown and pieces' colors are variations of parent's color, attempting to show that there's a relationship between the slice from the first chart and the pie chart with the details. Unfortunately, it wasn't possible to use similar lines like in Excel to show the relation between the two sections. 

Pie of Pie in Power BI

Instead of a pie chart, one can use a donut, like in the original representation. Even if the donut uses a smaller area for representation, in theory the pie chart offers a better basis for comparisons, at least in theory. Stacked column charts can be used as well (see C), however one loses the certainty that the pieces must add up to 100%. Further limitations can appear when one wants to achieve more with the visualizations.

Custom charts can be used as well. The pie chart coming from xViz (see D) allows to increase the size of a pie piece by using another radius, technique which could be used to highlight the piece represented in the second chart. Frankly, sunburst diagrams (see E) are better at representing the parent to child proportions, where the same color encoding has been used. Unfortunately, the more information is shown, the more loaded the visualization seems to be.

Pie of Pie Alternatives in Power BI I

A treemap can prove to be a better representation alternative because it encodes proportions in a unitary way, much like pie charts do, though it takes more space if one wants to make the labels visible. Radial charts (see G) and Aster plots (see I) can be occasionally better choices, especially because they use less space as they display only the main categories. A second diagram chart can be used to display the subcategories, much like in A and B. Sankey charts (see H) can be used as well, even if they don't allow representing any quantitative values unless one encodes them directly in the labels. 

Pie of Pie Alternatives in Power BI II

When one dives into the world of diagrams and goes behind the still limited representational choices provided by the standard tools, one can be surprised by the additional representational choices. However, their appropriateness should be considered against readers' skillset to read and interpret them! Frankly, the alternatives considered above could be a better choice when they will reach a representational maturity. 

Many thanks to Christopher Chin, who in his weekly post on data visualization blunders, suggested the examples used as basis for this post (see [1])!

Previous Post <<||>> Next Post

References:

[1] LinkedIn (2024) Christopher Chin's post (link)

📊Graphical Representation: Graphics We Live By (Part IX: Word Clouds in Power BI)

Graphical Representation Series

A word cloud (aka tag cloud) is a visual representation of textual data in the form of a cloud - a mass of words in which each word is shown with a different font size and/or color based on its frequency, significance or categorization in the dataset considered. It is used to depict keyword metadata on websites, to visualize free form text or the frequency of specific values within a categorical dimension, respectively to navigate the same. 

Words can be categorized as single or compounded, where special characters like hyphen can be used. A tag is a special type of a word, usually a single word. One can use different direction or arrangement for displaying each word, independently of whether the value is numerical or alphanumerical. Word clouds are usually not sorted, even if the values could be sorted using a spiraled arrangement, which offers and easier way to navigate and compare the data.

Most of the representations are based on words' frequency, though occasionally the frequency is considered against a background corpus (e.g. Wikipedia). The use of tags as categorization methods for content items is seldom done, though needs to be considered as well. 

It makes sense to use word clouds only with categorical data (see below) for which the chances of multiple occurrences is high. Numerical values (see A & D) can be displayed as well when their range is narrow. Moreover, when the number of distinct values is high, one can consider only the top N values. Continuous data may be challenging to represent, though occasionally they can be represented as well, especially when reducing the precision

Word clouds allow to see at a glance what values are available and can be used as an alternative to choropleth maps for filtering and navigating the data. They aren't good for precise comparisons, though further information can be provided in the tooltip. 

In Power BI there are currently two visuals that allow to display word clouds - from Microsoft, respectively Powerviz, which was added recently (see Jun-2024 release [2]). They provide similar functionality, though Powerviz's visual offers more flexibility in what concerns the word options (case, styling, delimiters) direction, shapes (displaying the values within a form), ranking (top vs bottom), exclusion rules and formational formatting. It uses also a radial arrangement, which allows to select or exclude a set of values via the lasso functionality (see E). 

Word Clouds

Previous Post <<||>> Next Post

References:
[1] Wikipedia (2024) Tag cloud (link)
[2] Microsoft Power BI Blog (2004) Power BI June 2024 Feature Summary (link)

📊Graphical Representation: Graphics We Live By (Part VIII: List of Items in Power BI)

Graphical Representation Series

Introduction

There are situations in which one needs to visualize only the rating, other values, or ranking of a list of items (e.g. shopping cart, survey items) on a scale (e.g. 1 to 100, 1 to 10) for a given dimension (e.g. country, department). Besides tables, in Power BI there are 3 main visuals that can be used for this purpose: the clustered bar chart, the line chart (aka line graph), respectively the slopegraph:

Main Display Methods

Main Display Methods

For a small list of items and dimension values probably the best choice would be to use a clustered bar chart (see A). If the chart is big enough, one can display also the values as above. However, the more items in the list, respectively values in the dimension, the more space is needed. One can maybe focus then only on a subset of items from the list (e.g. by grouping several items under a category), respectively choose which dimension values to consider. Another important downside of this method is that one needs to remember the color encodings. 

This downside applies also to the next method - the use of a line chart (see B) with categorical data, however applying labels to each line simplifies its navigation and decoding. With line charts the audience can directly see the order of the items, the local and general trends. Moreover, a line chart can better scale with the number of items and dimension values.

The third option (see C), the slopegraph, looks like a line chart though it focuses only on two dimension values (points) and categorizes the line as "down" (downward slope), "neutral" (no change) and "up" (upward slope). For this purpose, one can use parameters fields with measures. Unfortunately, the slopegraph implementation is pretty basic and the labels overlap which makes the graph more difficult to read. Probably, with the new set of changes planned by Microsoft, the use of conditional formatting of lines would allow to implement slope graphs with line charts, creating thus a mix between (B) and (C).

This is one of the cases in which the Y-axis (see B and C) could be broken and start with the meaningful values. 

Table Based Displays

Especially when combined with color encodings (see C & G) to create heatmap-like displays or sparklines (see E), tables can provide an alternative navigation of the same data. The color encodings allow to identify the areas of focus (low, average, or high values), while the sparklines allow to show inline the trends. Ideally, it should be possible to combine the two displays.  

Table Displays and the Aster Plot

One can vary the use of tables. For example, one can display only the deviations from one of the data series (see F), where the values for the other countries are based on AUS. In (G), with the help of visual calculations one can also display values' ranking. 

Pie Charts

Pie charts and their variations appear nowadays almost everywhere. The Aster plot is a variation of the pie charts in which the values are encoded in the height of the pieces. This method was considered because the data used above were encoded in 4 similar plots. Unfortunately, the settings available in Power BI are quite basic - it's not possible to use gradient colors or link the labels as below:

Source Data as Aster Plots

Sankey Diagram

A Sankey diagram is a data visualization method that emphasizes the flow or change from one state (the source) to another (the destination). In theory it could be used to map the items to the dimensions and encode the values in the width of the lines (see I). Unfortunately, the diagram becomes challenging to read because all the lines and most of the labels intersect. Probably this could be solved with more flexible formatting and a rework of the algorithm used for the display of the labels (e.g. align the labels for AUS to the left, while the ones for CAN to the right).

Sankey Diagram

Data Preparation

A variation of the above image with the Aster Plots which contains only the plots was used in ChatGPT to generate the basis data as a table via the following prompts:

• retrieve the labels from the four charts by country and value in a table
• consolidate the values in a matrix table by label country and value

The first step generated 4 tables, which were consolidated in a matrix table in the second step. Frankly, the data generated in the first step should have been enough because using the matrix table required an additional step in DAX.

Here is the data imported in Power BI as the Industries query:

let
    Source = #table({"Label","Australia","Canada","U.S.","Japan"}
, {
 {"Credit card","67","64","66","68"}
, {"Online retail","55","57","48","53"}
, {"Banking","58","53","57","48"}
, {"Mobile phone","62","55","44","48"}
, {"Social media","74","72","62","47"}
, {"Search engine","66","64","56","42"}
, {"Government","52","52","58","39"}
, {"Health insurance","44","48","50","36"}
, {"Media","52","50","39","23"}
, {"Retail store","44","40","33","23"}
, {"Car manufacturing","29","29","26","20"}
, {"Airline/hotel","35","37","29","16"}
, {"Branded manufacturing","36","33","25","16"}
, {"Loyalty program","45","41","32","12"}
, {"Cable","40","39","29","9"}
}
),
    #"Changed Types" = Table.TransformColumnTypes(Source,{{"Australia", Int64.Type}, {"Canada", Int64.Type}, {"U.S.", Number.Type}, {"Japan", Number.Type}})
in
    #"Changed Types"

Transforming (unpivoting) the matrix to a table with the values by country:

IndustriesT = UNION (
    SUMMARIZECOLUMNS(
     Industries[Label]
     , Industries[Australia]
     , "Country", "Australia"
    )
    , SUMMARIZECOLUMNS(
     Industries[Label]
     , Industries[Canada]
     , "Country", "Canada"
    )
    , SUMMARIZECOLUMNS(
     Industries[Label]
     , Industries[U.S.]
     , "Country", "U.S."
    )
    ,  SUMMARIZECOLUMNS(
     Industries[Label]
     , Industries[Japan]
     , "Country", "Japan"
    )
)

Notes:

The slopechart from MAQ Software requires several R language libraries to be installed (see how to install the R language and optionally the RStudio). Run the following scripts, then reopen Power BI Desktop and enable running visual's scripts.

install.packages("XML")
install.packages("htmlwidgets")
install.packages("ggplot2")
install.packages("plotly")

Happy (de)coding!

Previous Post <<||>> Next Post

📊Graphical Representation: Graphics We Live By (Part VII: Reading a Conversion Rates Chart with ChatGPT and Copilot)

Graphical Representation Series

One of the areas where ChatGPT, Copilot and other similar AI-based chatbots can help is in summarizing a chart saved as image. Ideally, the chatbots should be able also to approximate the points from the chart as well (an image is made of pixels and thus areas should be easy to delimit). So, I was wondering how far the chatbots can be used for these purposes. I used first an image copied from the web, though I realized that not all visual elements could be read (e.g. Copilot had issues retrieving the values for some months) and I had no basis data for comparisons to identify how big the deviations are. 

So, I created a chart in Power BI based on the below chart (see original data):

Conversion Rates Dual Axes Chart

Here's the output based on Copilot over several attempts:

	Original data			First attempt		Second attempt		Third attempt		Fourth attempt
Sorting	Month	Conv.	Conv. Rate	Conv.	Conv. Rate	Conv.	Conv. Rate	Conv.	Conv. Rate	Conv.	Conv. Rate
1	Jul	8	4	10	1	10	1	8	4	8	4
2	Aug	280	16	275	15	275	15	275	18	275	18
3	Sep	100	13	225	12	225	10	225	12	225	12
4	Oct	280	14	275	12	275	11	275	11	275	11
5	Nov	90	4	75	5	75	6	75	6	75	6
6	Dec	85	3.5	100	5	100	5	100	5	100	5
7	Jan	70	4.5	50	3	50	3	50	4	50	4
8	Feb	30	1.5	50	3	25	2	50	2.5	50	2.5
9	Mar	70	4	25	1	50	2.5	25	1.5	25	1.5
10	Apr	185	11	200	10	200	10	200	10	200	10
11	May	25	3.5	50	4	50	4	50	3.5	50	3.5
12	Jun	195	4	225	10	225	10	225	11	195	4

Please note that same values were repeated to identify whether the chatbot is able to identify the resemblance correctly. 

Based on the data several tests were made with ChatGPT4o, where the following prompts were given:

1. generate the data behind the chart as excel list (see 'First attempt' in the table)
2. generate the data behind the chart as excel list where blue is represented on the left axis and orange on the right axis (see 'Second attempt');
3. reevaluate the values from the previous output considering that the values for Jul are 8, respectively 4% (see 'Third attempt');
4. reevaluate the values from the previous output considering that the values for Jun are 195, respectively 4% (see 'Fourth attempt');
5. generate the data behind the dual axis chart as list from the last conversation;
6. summarize the chart by revealing trends and other statistics.

From attempt to attempt, the values get closer to the original data, corrections being made, though there are points in the final output that diverge considerably. ChatGPT does generate a table which can be easily copied to MS Excel, and even gives some instructions on how to generate a dual axis chart. 

It's great that one can use data between sessions and improve maybe the output. ChatGPT's summary is better than expected and it leverages the latest corrections. However, the estimations need to improve so that the inferences can be correct. 

Here's the output based on the same prompts in Copilot:

	Original data			First attempt		Second attempt		Third attempt		Fourth attempt
Sorting	Month	Conv.	Conv. Rate	Conv.	Conv. Rate	Conv.	Conv. Rate	Conv.	Conv. Rate	Conv.	Conv. Rate
1	Jul	8	4	300	18	300	18	8	4	8	4
2	Aug	280	16	250	16	250	16	250	16	250	16
3	Sep	100	13	200	14	200	14	200	14	200	14
4	Oct	280	14	150	12	150	12	150	12	150	12
5	Nov	90	4	100	10	100	10	100	10	100	10
6	Dec	85	3.5	50	8	50	8	50	8	50	8
7	Jan	70	4.5	300	18	300	18	300	18	300	18
8	Feb	30	1.5	250	16	250	16	250	16	250	16
9	Mar	70	4	200	14	200	14	200	14	200	14
10	Apr	185	11	150	12	150	12	150	12	150	12
11	May	25	3.5	100	10	100	10	100	10	100	10
12	Jun	195	4	50	8	50	8	50	8	195	4

Copilot's estimations are higher than the ones made by ChatGPT and deviate more from the original data. No reevaluations are done between prompts for the other values. The summary provides information that can be used to complement ChatGPT's output. 

Overall, ChatGPT seems to perform better than Copilot, at least for this example (though we might talk here about different "generations"). Unfortunately, given that the estimations provided by both chatbots deviate considerably from the expectation, the output needs to be revised and corrected, which decreases the usability of such chatbots. In fact, one can use them to generate an initial set of data and correct then the deviations.

The outputs of other chatbots like Google's Gemini or Claude-3-Haiku (via Poe) can't be compared with the ones from ChatGPT or Copilot yet. Claude-3-Haiku does provide estimated values (even with comma), though they deviate considerably from the original data. 

It would be interesting to test how other charts and plots are processed by chatbots, respectively whether the various visual elements (e.g. gridlines, ticks, markers) make a difference.

Previous Post <<||>> Next Post