Assignment 2: Burtin’s Antibiotic Data

[venkatrajam]

In the post-World War II world, antibiotics were called “wonder drugs,” for they provided quick and easy cures for what had previously been intractable diseases. Data were being gathered to aid in learning which drug worked best for which bacterial infection. Being able to see the structure of drug performance from outcome data was an enormous aid for practitioners and scientists alike. In the fall of 1951, Burtin published a graph showing the performance of the three most popular antibiotics on 16 bacteria.

The data used in his display are shown in attached Excel file. The entries of the table are the minimum inhibitory concentration (MIC), a measure of the effectiveness of the antibiotic. The MIC represents the concentration of antibiotic required to prevent growth in vitro. The covariate “gram staining” describes the reaction of the bacteria to Gram staining. Gram-positive bacteria are those that are stained dark blue or violet; whereas, Gram-negative bacteria do not react that way.

Present a graphical representation of these data and an accompanying written description of the graph.

(this assignment is a rerun of the Chance magazine's contest on the 100th birth anniversary of designer Will Burtin who published the first visualization of the data in 1951. Contest winners' submissions are here.)

Check out the last batch's submissions for the assignment.

[pujan2000]

Name - Pujan Chitalia

Aim: To understand the most effective antibiotic and the dosage requirement to cure a patient suffering from any of the given bacteria.

Started with plotting proportions of three antibiotics required for each Bacteria. In this graph some proportions are so low that it is not visible. Also, this type of bar graph represents that we need some combination of 3 antibiotics to diagnose/kill any given Bacteria. The graph type was bar graph since the data is discrete.

In this graph, I have used ln(minimum inhibitory concentration) and changed the axis.

For better representation purposes, I plotted the graph using logarithmic scale and added a data table at the end. • The MIC is basically the concentration of antibiotic required to kill the bacteria. So, lower the MIC, better is the performance of the antibiotic. • The MIC is plotted on the x axis and the bacteria is plotted on the y axis. • I have separated the plots on the basis of gram stain in order to check the effectiveness of MIC on the bacteria in case of negative or positive gram stain. • Penicilin seems to be extremely effective on gram positive bacteria. • For gram negative bacteria, Neomycin seems to be most effective. Also, Neomycin’s performance for gram positive bacteria is decent. • Performance of Streptomycin as an antibiotic is not good in gram positive or gram negative bacteria.

[Stuti-018]

Name- Stuti Pasricha

This is the Visual representation of Burtin's Antibiotics data. Three antibiotics' information against 16 germs is revealed by the data set. Please take note that a certain antibiotic is regarded to be more effective the lower the value. I have normalized the data by taking log of each column at base 10. But, that may give negative values and so we added a number (4) to the result to shift all the values at positive axis. I grouped all three medications, plotted their efficacy, and stacked all 16 bacteria on the x axis to enable a visual comparison of all three drugs. By doing so, it would be possible to compare all the antibiotics both in relation to one specific bacteria and to the other 15 microorganisms. I also encode the gram nature by spearing the bacteria as per their category and color encoding is there to differentiate between negative and positive gram stain. Though neomycin was the most successful of the three, I discovered an intriguing pattern: the results were varied depending on whether the samples were Gram Staining Positive or Gram Staining Negative. I discovered that penicillin was ineffective in killing gramme negative bacteria, but was incredibly successful at killing gramme positive bacteria.

[kirubanath]

Name: Kirubananth Sankar

Objective: To give an easy and faster way for the medical professional to check the chart and give the appropriate dosage of antibiotic

Description:

1. The chart is divided into two parts based on Negative gram stain and Positive Gram stain.
2. The x-axis is a weight-adjusted logarithmic scale. (weight adjusted to improve visibility)
3. The y-axis has the types of bacteria.
4. Three different colors have been allocated to the three different antibiotics we have.
5. The bold annotation denotes the appropriate shot that needs to be administered for a given bacteria.

Rationale:

1. The main reason is choosing this was because I was thinking from the perspective of a medical practitioner. I wanted to give a way for them to access the information quickly
2. The scale was adjusted for the same purpose to create more visibility. But the annotations which are in the bold, which are what they will access are accurate.
3. Bar chart was chosen rather than any other graph, to reduce the complexity and to order it systematically in alphabetic order, so it becomes easy for them to search.

[Beastwham]

Name: Prashant Kapoor

Description:

• The chart has been divided into two parts based on gram stain.
• The values on these graphs have been adjusted with logarithmic scaling (base 10).
• Formula: 5 - log10(x) where x is the value.
• The outermost value for any bacteria is the drug that uses the least amount of drug, and the innermost value uses the highest amount of drug.
• Doctors can recommend the antibiotic farthest from the centre for each bacteria.
• Interestingly enough, for bacteria with positive gram stain, streptomycin requires the highest dosage and for bacteria with negative gram stain, penicillin requires the highest dosage.

Note: I'd like to mention that while the above graphs do a good job of representing the information, if we were to create a simple instructional infographic for doctors then it'd have to be much simpler.

I tried to look for clusters by plotting them on scatterplot and tried to create an infographic that would serve this purpose.

Here's the infographic:

Here's the link for the jupyter notebook in which I plotted the data: https://colab.research.google.com/drive/16-5fRceo8K3Z_h564yw92T0_SJIFKtMz?usp=sharing

[vanshu2k]

Name: Vanshu Saini

Platform Used: Tableau

A different Graph link: https://drive.google.com/drive/folders/1yv5KIcEtBh34qP1muhBi5D4rlBInlS4m?usp=sharing

Description:

-The graph is divided between positive & negative gram staining. -There are three different colors used for each Penicillin, Streptomycin, and Neomycin. -The x-axis on the graph has been adjusted and normalized to the logarithmic scaling of base 10. -Bars are used to show an accurate representation of the data and also to aid in comparing the effects of various antibiotics. -The y-axis has been labeled with the names of different bacteria.

[subandwho]

Name: Subandhu

Platform : Tableau Visualization used: Line Chart.

Description:

• The given chart segregates bacterium into groups based on their Gram Staining.
• The peaks of each line chart represents the efficacy of given antibiotic vs the bacterium on y axis.
• The efficiency values have been scaled using logarithmic scaling as that allows all values to be comparable and easy to refer.
• A line chart serves the purpose of conveying all relevant information and immediately assist when someone wishes to check for efficiency against a particular bacteria.

[vivek6311]

Vivek Kumar, TLP-23

Description

1. It is easy to visualise when you divide the graph into two parts i.e. with positive and negative gram stains.
2. Penicillin is used in high amounts when gram staining is negative.
3. Penicillin is used in Low amounts when gram staining is positive.
4. While the other two components are used in high amounts when gram staining is positive.
5. The data representation in the form of a column chart is easy to visualise as it also shows the percentage of different components used to represent this.

[gaurav639]

Name - Gaurav Singh

Platform Used: Tableau

Communication Goals of my Visualization:

• My visualization allows the viewer to make judgements about which antibiotic they believe is both overall most effective, and most effective in certain cases against specific bacteria.

Design rational:

• I first tried scatterplot (each drug’s effectiveness per bacteria) in order to compare each drug to one another, but it was too messy.
• I finally decided to separate the three antibiotics to demonstrate their overall characteristics. On the X-axis I listed MIC (log transformed) to address the vast distribution in MIC scores. On the Y-axis, for every antibiotic, I listed all the bacteria.
• First, I tried bars for plotting, but it took up so much of space then I finally plotted the MIC scores using square marks against bacteria. This also allows the viewer to judge each drug against one another. Also, positive and negative Gram staining is clearly shown through different color.
• I used an odd approach that drug effectiveness is decreasing as you read to the right, but the description provided at the right side of the chart make it easy to understand.

[Suvrojyoti-Paul]

Name- Suvrojyoti Paul

Visualization used: Bar Chart.

Description:

There are two charts, one is for the positive gram straining and negative gram straining. The x-axis shows percentage proportion Penicillin is shown by green; streptomycin is shown by black, and neomycin was shown by blue, Each bar shows what percentage of penicillin, streptomycin and neomycin needed for each kind of bacteria. Penicillin is more dominating for negative straining Streptomycin is more dominating in positive straining.

Design rational:

I chose this structure because I wanted the user to visualize the combination of antibiotics (penicillin, streptomycin and neomycin) needed for each bacteria straining and what is dominating more in negative and positive gram straining.

[anchitna]

Name - Anchit Narayan

Platform : Tableau Visualisation : Bar Chart Description :

• The vertical axis shows the value of minimum inhibitory concentration (MIC) which has been converted to logarithmic scale to make it easier to compare.
• The lower the value on the vertical axis, the more effective it is.
• The higher the value on the vertical axis, the less effective it is.
• The horizontal axis has been broken down into different bacterias, each having it's mapped MIC with Neomycin, Penicilin and Streptomycin.
• Two different charts for positive and negative gram staining makes it easier to differentiate and understand the chart.

[PratProduct]

Name: Pratyay Prakhar

The following 3 visualisations have been created to answer some questions by drawing inferences from 'Brutin's antibiotic data'.

Question 1: How does one antibiotic compare to another in terms of efficacy ?

Design rationale:

• A unified view of both the virus strains is created for easy comparison and to make the scale for comparison common. The values for one strain varied from 0-870 while it varied from 0-100 for the other. Thus, usage of the same scale makes the comparison easier. Had the graphs for both strains been separate, the logarithmic scale for each based on the data values would have been different which would have made the comparison tougher.
• The y-axis represents the MIC and the x-axis represents the Bacteria.
• No lines have been created between the points because the data is discreet. So, the lines were not created to prevent users from thinking that the data was continuous in nature.
• Vertical grid lines have been added to make it easier for the user to compare the 3 data points for a bacteria. Not having the grids increased the load on the user to trace the data points which lie along the same vertical line for a bacteria.
• A simple colour-coded line is used to distinguish between the 2 strains of the virus. An arrow line was not used because that would have indicated the existence of some order in the axis.
• The y-axis has been converted to a logarithmic scale to make the graph more legible. If the graph y-axis was not concerted into a logarithmic scale from a linear one, the extreme values-870 and .001 would have resulted in a strange distribution impacted by the outlier points.

Question 2: Which is the most effective antibiotic for each bacteria?

Design Rationale:

• Since the most effective antibiotic has the least MIC value, a derived column is created which contains the most effective bacteria for each bacteria.
• A pictogram is created with the symbols/ pictures of the most effective antibiotic for each bacteria. A lot of pamphlets that doctors use make use of pictograms for easy readability.
• Another choice could have been to create a table of bacteria and the respective most effective antibiotic but that would have made it tougher for the user to identify patterns and identify the dominant antibiotic for a particular strain.
• The picture used is that of a capsule because it closely represents an antibiotic medicine.
• The colour coding used is in sync with the colour coding used in the previous visualisation in order to avoid any confusion and conflict within the mental model.

Question 3: Which antibiotic is effective if only the gram strain of the bacteria is known?

Design Rationale:

• Since the gram strain of the bacteria can be determined within a short time before even determining what bacteria it is, a doughnut chart is used to show the efficacy of the antibiotics against both the viral strains.

[anav99]

Name: Anav Bhatia

Description:

1. X-axis indicates the Minimum Inhibitory Concentration(MIC)
2. Y-axis indicates the Bacterial Infection
3. The graph has been divided into 2 parts for Gram Positive and Gram negative Staining
4. Logarithmic scale has been used in x-axis.
5. The whole display has been divided into two parts- one with the graph, the other with the legends and some important information.

I choose this style because it represents the information in a minimalistic way and appeals to the eye.

[ghost]

Name - Savyasachi Pandey

Platform - Tableau Visualization - Horizontal Bar Chart

Rationale for bar chart: I chose the horizontal bar chart as in this the labels are easier to display. It shows each data category in a frequency distribution, also it estimates the key values at a glance.

DESCRIPTION:

1. X Axis depicts the Minimum Inhibitory Concentration (MIC), which has been converted using logarithmic scale to make it easier to compare as the values were varied.
2. As the value on the horizontal axis goes down, the effectiveness increases and vice versa.
3. The vertical axis has been differentiated into different bacterias, each showing its MIC with Neomycin, Penicillin and Streptomycin.
4. Two different graph for positive and negative gram straining makes it easier to read and understand the graph as well as makes the mind mapping easy.
5. The Visualization above shows that the most effective antibiotic for gram-positive bacteria is Neomycin, whereas for gram-negative, it is Pencillin.
6. Simple colour coding technique is used with the help of 3 colours, i.e. Orange for Neomycin, Purple for Pencillin and Red for Streptomycin.

[Shiv907]

Description:

1) Log (base 10) has been applied to all the numerical values of the data. This is done to reduce the variability and normalize the data. 2) The data has been divided into two categories: Positive gram staining and Negative gram staining. 3) It is clear from the data that Penicillin is extremely ineffective as an antibiotic for gram-negative bacteria whereas it does fairly against gram-positive. The indicator used for this inference is the presence of blue color in both charts on the left side and on the right side.

[29pranav]

Description of the Graph:

1) The data has been divided into two segments: Positive staining and Negative staining. 2) To reduce the large variance inside the data, logarithm (log to the base 10) has been taken. Otherwise, it was difficult to include such a large range (0.1 to 870) in a single chart. 3) Each bacterium's MIC has been compared against all 3 drugs in each of the two charts (Positive stain as well as Negative stain)

Conclusion:

1) The performance of Penicillin is the best amongst all three against gram positive bacteria. 2) The performance of Neomycin is the best amongst all three against gram negative bacteria.

** The drug having least value of MIC for a particular bacterium is the best performing.

[ShivayNagpal]

Name - Shivay Nagpal Graph used- Radial column chart

Since our data only had $16$ categories of bacteria, I chose a radial column chart to plot the data as $16 \bigtimes 3$ types of antibiotics $= 48$ data points, which can be easily encoded in a radial graph and still remain legible. I initially chose a logarithmic scale ranging from $10^{-3}$ to $10^3$, since all of our data was captured in this range. The problem with this range was that some data points ended up having zero height, which defeated the entire purpose of creating the infographic since a reader will not be able to distinguish the most effective antibiotic (the shortest height bar for the particular bacteria). To fix this issue, I changed the radial axis to vary from $10^{-4}$ to $10^3$, which ensured a minimum height for all the bars.

[Aayush4396]

1. I have converted the Minimum inhibitory concentrations (MIC) values and taken the logarithmic values of MIC as the order of magnitude is huge for the data. I get negative values and the min value is -3, so I've added a constant of 5 to all the log values: Converted MIC values => log(MIC) + 5
2. After having all the values as positive I took the percentage (antibiotic-wise). The y-axis consists of the bacteria type and the x-axis consisted of the converted MIC values. There are 2 separate plots of gram-negative and gram-positive.
3. Now the effectiveness is decided if the MIC value is low, hence I have highlighted the lowest converted MIC value for each bacteria.
4. The questions that the plot answers are: a. Which antibiotic to take given the MIC values? b. Comparison of effectiveness of antibiotics when MIC values are given.

[SonicShotGaming]

Name: Priyam Shah

So what I aimed to do with this visualization is to first split the bacteria into Gram Positive and Gram Negative. This would make it easier for the viewer to easily access the information that they need. What the graphs are depicting is the amount of each antibiotic a bacteria needs to reach MIC, with the lesser amount being the better antibiotic. This means that the bars which extends the most towards the left side is the most effective antibiotic for that bacteria.

I chose to go with a dark theme with very vibrant colours so that the most important information would pop out, contrasting with the dark background. I also chose a bar graph like this because it would be very easy for a viewer to see which bar extends the most to the left and then chose the best antibiotic for the bacteria. In a life or death situation, the viewer might not have the time to process and figure out which antibiotic might save the life of the patient, and having the graphs be so vibrant will draw their eyes to the bar, and finding the leftmost bar is also very intuitive.

The questions that my visualisation answer are:

1) What are the MIC values for antibiotics for certain bacteria? 2) Which antibiotic is the best for each specific bacteria? 3) Which antibiotic can be used for the overall Gram Positive and Gram negative group.

[GaneshBShelke]

Name: Ganesh Shelke

• I have used bar chart which effectively shows the comparison of the Minimum Inhibitory Concentration (MIC) of the antibiotics required for different bacteria, when plotted on the log scale.

• The least value on the bar graph represents that the antibiotic is most effective for that particular bacterium.

• I have made two separate charts for positive and negative gram staining to make it easy to differentiate between these two categories.

• Bar graphs are easy to comprehend for this type of data and don’t take much cognitive effort to understand the insights.

[vrk7]

Name: Vysakh R K Platform used: Power BI Link to Power BI: Performance of three most prescribed antibiotics

The principles which I have taken in mind when coming up with this graph is:

1. Minimalistic design
2. Easier for the users to comprehend the graph
3. Using a line graph to come to an inference, rather than using a complicated graph.

Description: Data from Burtin's paper published in the year is used to create the above graph, which shows the performance of the three most prescribed antibiotics on 16 bacteria. I have used the type of bacteria on the Y-axis and Minimum Inhibitory Concentration (MIC) on the X-axis on a log scale as the scaling factor so that large values can be normalized. My graph illustrates how vaccines affect different types of bacteria from the dataset. Taking this approach was my way of illustrating how each antibiotic contributes to a particular type of bacteria while also showing generalizations. Also, the maximum and minimum line (dashed lines) is mentioned in the graph to show the effectiveness of the drug.

Results: The most effective vaccine for most of the Positive Gram Straining bacteria: Neomycin The most effective vaccine for most of the Negative Gram Straining bacteria: Penicillin Streptomycin is not very effective against Positive gram nor Negative gram-straining bacteria.

[AdarshGowda03]

Name: Adarsh Gowda Tool used: Tableau

Description

1. I used a bar graph because it allows me to easily compare different sets of data among different groups. The Minimum Inhibitory Concentration (MIC) of antibiotics required for different bacteria is compared here.
2. The graph design is minimalistic because it includes 'Gram Staining' in the same graph that is differentiated by two colours, making it easier for the user to analyze the data in a single graph.
3. I used a logarithmic scale for effective comparison the data was skewed due to large values.
4. When compared to the concentration required for active growth, we find that Neomycin is more effective for negative gram staining.
5. Except for two bacteria, we found penicillin to be more effective against most bacteria when stained with positive gram staining.

[promwonka]

Name : Pramod Krishna Title: Effectiveness of antibiotic wrt gram strain

Description:

1. I segregated the positive and negative gram strain into 2 sections of the graphs with 2 different colours.
2. Also labels have been given for clear unerstanding.
3. The x-axis represents the MIC and the y-axis represents the Bacteria.

[itsnitiz]

Submitted by Nitish Mallick

Description:

• I have showcased everything in one chart. Its pretty much self explanatory

• I could have adjusted the values by log, etc. but again it will not serve the purpose as the user has to recalibrate each figure.

• I think this chart serves every purpose for doctors and everyday people.

[BhanuN1997]

Name: Bhanu Nalluri Platform: Excel

Technical limitation(Edited on the 20th): I couldn't increase the text size and contrast due to technical limitations (spoke with the professor on this in class). But the underlying idea is to present everything in one chart and differentiate based on the family strain

Research Methodology: I read through a few articles and research papers to understand the concept of strains and the significance of MIC values.

Findings and implementations:

Good vs Bad: We should not assume that lower MIC values are better. The ideal values depend on the ingredients, and complexity of manufacturing, and the reactions of the API. So, I have avoided mentioning what's good and what's bad while benchmarking MIC values

Units used: I have used µg/mL for measuring MIC concentration. MIC is generally represented in µg/mL (1 µg = 1000 mg)

Importance of MIC values: A slight change in MIC numbers even at the decimal level may impact the effectiveness of the drugs. So I felt it's important to represent the raw data. Plotting a histogram based on ranges along with raw data will give a better picture to causal readers

Users: The Viz. is targeted at both normal users and users from the healthcare industry. I have mentioned the raw numbers below the Histogram for medical users as they prefer to know the exact MIC values of each drug while benchmarking. I have benchmarked the MIC values based on ranges in the form of a histogram to provide a snapshot for normal users.

Constraints: The given data is relatively hard to represent in one picture. I felt it's important to represent the whole data in one picture, so it will be easy for the user to benchmark different values just by looking at one picture

Aesthetics: To avoid unnecessary distractions for the readers I have avoided the usage of fancy designs/pics and presented the data in a minimalistic format.

[monikrish2698]

Name: Monish Krishnan Tool Used: Power Bi

I initially thought of representing the data using a stacked bubble chart or vertically stacked bar chart. While designing them, I realized it could lead end users to misinterpret bubbles and stacked bars. I wanted to design a visualization that should be easy for viewvers' eyes but also help them gain some insights. To achieve my viz-representation goals, I finally chose vertically bar charts with the different bacteria on the y-axis, MIC values on the x-axis, and the antibiotics on the legend. I differentiated them by their gram stain test output and placed them side to side to make comparison easy. It was interesting to see the increased concentration of Penicillin in negative gram stain and almost zero concentration of Penicillin in positive gram stain.

My Design principles:

1. Easy to viewers' eyes
2. Insights
3. Minimalistic design

[Pragyawasan]

Chart type- Bar chart Tool- Tableau

• A divided bar graph is chosen here as I wanted both the positive and negative gram stains on the same graph yet separated so that they are easier to compare.

• All values for minimum inhibitory concentration (MIC) have been normalized to a logarithmic scale of base 10.

• The positive values of minimum inhibitory concentration (MIC) are represented on top and negative values at the bottom as we are used to reading graphs with the positive axis on top and negative at the bottom.

• It has been sorted alphabetically by the bacteria names as an assessor would be looking for a particular bacterium and once they find it, they can see whether the gram stain is positive or negative and what are the MICs of the different antibiotics.

• The question that my visualisation is trying to answer that for a particular bacterium, which antibiotic is the best choice and hence all the above steps are done in order to make it simpler and quicker for assessor to view the data and quickly choose from it.

[ada-nai]

Name: Adarsh Nair Tool used: Matplotlib, Seaborn (Python3)

• The first two visualizations depicts the effectiveness of the three antibiotics (Penicilin, Streptomycin and Neomycin) against Gram Negative Bacteria and Gram Positive Bacteria using horizontal grouped bar charts respectively. This type of visualization was preferred as it provides a simple, yet effective way to compare the MIC values of all three antibiotics
• The antibiotic with the least Minimum Inhibitory Concentration (MIC) value among the three is the preferred one to be administered against the specific bacteria/virus mentioned in the X-axis
• The MIC values are condensed by logarithmically scaling the range
• The third figure indicates the general effectiveness of antibiotics against bacteria/viruses using a stacked bar plot. It is observed that Neomycin requires a lower MIC for both gram positive and gram negative bacteria/viruses. Besides, Penicilin requires low MIC for gram positive organisms and Streptomycin requires low MIC for gram negative organisms

[walia-muskaan007]

Name - Muskaan Walia Chart Type - Bar Chart Tool - Tableau

I have used a vertical bar graph to differentiate between the efficacy of the antibiotics on the Gram Staining results to find the minimum inhibitory concentration (MIC) on the logarithmic scale.

• Each vertical column represents the bacteria.
  • For the Gram-Negative results, we find that the lowest MIC is that of Neomycin. Hence, it is the most effective antibiotic against the Gram-negative bacteria.
  • For the Gram-Positive results, we cannot pinpoint any one of the antibiotics. There is a clash between the lowest MIC - Neomycin and Penicillin. Still, the use of Penicillin seems to be slighter greater. Hence, it is the most effective antibiotic against the Gram-positive bacteria.

[tirthatilakpani]

Antibiotic Data Set visualization.pdf Name- Tirtha Tilak Pani Important clarification- In the case of the Bacteria Proteus Vulgaris, we see that the MIC values for Streptomycin and Neomycin are the same. I have chosen Neomycin over Streptomycin in that case because Neomycin is the mode in the case of the second visualisation. While I feel in the initial stages, the distinction between Gram positive and Gram negative is essential, as we progress, I think that it does not have as much practical impact on data analysis as the three anti-biotics.

[ShashankPrasad0506]

Name: Shashank Prasad

The y-axis represents the categories of bacteria. The x-axis represents the MIC (Minimum Inhibitory Concentration) for each antibiotic on a log scale.

Bar chart is used to effectively show the comparison of the Minimum Inhibitory Concentration (MIC) of the antibiotics required for different bacteria. The scale is changed to log scale to make it easier for visualization. The antibiotic which has a lower value against a bacteria works better against that bacteria.

Two separate charts for positive and negative gram staining are shown for easy visualization and less clutter.

[Tamanna134]

Name: Tamanna Sahu

Platform used: Tableau Bar chart has been used to compare the result of three different antibiotics on each of the 16 bacteria. It's easy to compare which antibiotic has been more efficient by the length of the bar for that particular bacteria. On x-axis, MIC values have been normalized to logarithmic base 10. On y-axis, bacteria have been labelled and arranged in alphabetical order along with mention of positive/negative gram bacteria type.

[Harsss]

Burtin's Antibiotic Data

• Platform Used: Microsoft Excel
• Two Group Charts, One for Gram Negative and One for Gram-Positive.
• The Y axis is had the logarithmic value starting from 0.001 to 1000
• The reference used is at 1
• The one that is best had a crown on the top of it to represent that it's the best for usage

[abdksyed]

Name: Syed Abdul Khader Tool Used: Python - Seaborn

The y-axis represents the nominal data which are the categories of the bacteria. The x-axis represents the MIC(Minimum Inhibitory Concentration) for each antibiotic on a log scale.

For each bacteria, we have all three antibiotics plotted, with their values, we can pick the antibiotic as better for that bacteria which has a lower value. So the visualisation is made accordingly to help with that.

Also, the gram strain is binary, so it is visualised as the background colour and the bacteria's font colour where the positive is green and the negative is red to match with the user's mental model.

[kishlaykumar1995]

Name - Kishlay Kumar Chart Type - Bar Chart

Description

• Burton's antibiotic data measures the Minimum Inhabitory Concentration (MIC) and a higher concentration generally means lower efficiency of the drug.
• To address this unintuitive measure, I have modified the MIC data using a formula: $10+1/log(MIC)$
• The resultant values are then converted into percentages for each bacteria type which roughly represent the efficiency of the drug.
• The data is then displayed with Bacteria (nominal) as x-axis and Gram Strain (nominal) as y-axis seperated with hue representing the values for the 3 antibiotics.

[AyushiNM]

Name - Ayushi Mittal Title - Burtin's Antibiotic Data Visualization Platform used - Tableau

• In this visualization, I have used bar charts to display Minimum Inhibitory Concentration (MIC) of Penicilin, Neomycin and Streptomycin antibiotics on the logarithmic scale.

• We have two graphical representations, one for bacterias with Positive Gram Straining and the other for bacterias with Negative Gram Straining.

• The x-axis represents the bacterias and the y-axis represents MIC (Logarithmic scale).

• For each bacteria, the MIC for the antibiotics are represented by Red (neomycin), Cyan (penicilin) and Yellow (streptomycin) colored bars.

• I have used shade of blue to represent the distribution for Gram Positive Bacteria and shade of red to represent the distribution for Gram Negative Bacteria.

• For the Gram Positive Bacterias, Penicilin seems to be the most effective antibiotic, while for Gram Negative Bacterias, Neomycin seems to be most effective as these are used in least quanities in their respective cases.

• Streptomycin performs decently on Gram Negative Bacteria, but not so well on Gram Positive Bacteria.

• Penicilin doesn't perform well on the Gram Negative Bacteria and Necomycin doesn't bode well with Gram Positive Bacteria.

• The above conclusions can also be reviewed on the basis of the following Pie-charts that represent the proportion of usage of each antibiotics for the Gram Positive and Negative cases. The Pie-Charts use the same legend as the bar graph, i.e., antibiotics are represented by Red (neomycin), Cyan (penicilin) and Yellow (streptomycin) colors.

[AdityaPandey0901]

Visualizations

https://docs.google.com/drawings/d/1v6qNvKe_6EosIzH6Hos-t9_tcV7Lmab65hIgaokQP44/edit?usp=sharing

Explanation:

The larger goal of this visualization is to provide a holistic view of

a) How Gram +/- has an effect on the MCI's required for each drug.

b) The comparative MCIs of each drug to combat each bacterial agent.

To achieve this, I've opted for a set of 3 plots:

1) The first of these is a 3d autorotating plot, stored as a gif. This might stretch the definition of a "static visualization" (Note the definition we were given in class was no user interactivity, which gifs statisfy), but in doing so it provides a holistic representation of Gram and the role it plays in the efficacy of different drugs. For the researcher, this provides an excellent insight into an otherwise unseen trend.

2) The second and third, split each bacterial agent by Gram positivity, and provide multi-bar plots side by side for each. The bars are in high contrast colors, with the idea being that a doctor can easily look at the bacterial agent they want to combat and instantly weigh their options.

The axes chosen everywhere are logarithmic, and I've used plotly for the entire viz.

[HimanshuBhardwaj2398]

Name: Himanshu Bhardwaj Title: Comparing Effectiveness of Antibiotics Platform/Tool used: Jupyter Notebook(pandas, matplotlib )

Comments:

1.) Data contains MIC for three antibiotics(Penicilin, Streptomycin and Neomycin) for various gram- positive and gram-negatuve bacteria. The most effective drug for a particular bacteria would be the one, having least MIC value 2.) We can present the data in a form suitable to compare all three anti biotics for all kinds of given bacteria. The most intuitive choice for comparing seems to be a bar plot. For each bacteria, three bars represent the amount of MIC required and the drug corresponding to the bar with the least value will be the most effective.
3.) As MIC values vary a lot, values are modified using log(MIC)(base 10) to convert them into comparable ranges. Hence, wherever values are not present in the plot, it means a MIC of 1 4.) Looking at the visualization, a general trend can be observed: Penicilin is observed to be most effective(out of the given three drugs) against gram-positive bacteria and Neomycin is the most effective drug for gram-negative bacteria

[Kunalbhardwajj]

Name - Kunal Bhardwaj Title - Burtin's Antibiotic Data Visualization Platform used - Python Based on this plot, Neomycin is more effective against gram -ve bacteria than the other two. Based on this plot, Penicilin is more effective against gram +ve bacteria than the other two, here it seems like that the plotted line for Penicilin is passing through 0 but that is due to the normalizing effect, in actual its concentration needed to fight against gram +ve bacteria is very less in comparison with the other two, so the plotted line lies slightly above 0, not passing through 0.

Description:

• Since the dataset consists of MIC values of three antibiotics against 16 types of bacteria but for an antibiotic, the difference among values in MIC values against different bacteria was very large so to increase the uniformity in data and make it easier to navigate, I normalized the values for each antibiotic respectively.
• Note: MIC represents the concentration of antibiotic required to prevent growth in vitro, so the lower the value of MIC for a bacteria, the more effective that antibiotic is against that bacteria.
• To get plots that would be easily distinguishable for different bacteria and would be easier to interpret for the effectiveness of different antibiotics against different bacteria, I divided the dataset into two parts, one for gram -ve and another one for gram +ve, and used two separate plots for better representation.
• Note: Gram-positive bacteria are those that are stained dark blue or violet; whereas, Gram-negative bacteria do not react that way.
• My intention was to keep the plots intuitive yet informative so used line graphs only and since they are plotted against discrete values, I used dotted representations instead of continuous lines.

[asheesh4545]

Name - Asheesh Kumar Singh Chart Type - Bar Chart

Tools Used : Tableau

Description :

• The bacteria has been divided on the basis of two categories : gram positive and gram negative
• I have used horizontal bar graph since visualization was clear with this method

[Tejasvi337]

Name: Tejasvi Kumar Singh Tool: Plotly

The aim was to provide a descriptive visualisation of the performance of three antibiotics with minimalist aesthetics

Design rationale and features

• The particular layout was selected because -The data type was quantitative(discrete) therefore bar char/ scatter plot -It makes spotting the most effective antibiotic for a particular bacteria strain easier since it will be the first quantity on the left-hand side
• Some MIC values in the dataset were overlapping so to make this collision visible the opacity of the scatter plot was reduced
• Antibiotics are marked using different symbols to make the process of differentiation easier
• Gram-staining information is encoded using colour. Green colour is used for marking Gram-positive bacteria strains while red colour is used to signify Gram-negative bacteria strain
• On the y-axis, bacteria strains are categorised based on their Gram staining results and then sorted alphabetically. This is to make searching through different strain, by the user, easier

[SumanthSonnathi]

Name: Sumanth Sonnathi Title: Burtin's Antibiotic data analysis

Description: -> Data visualization happens to be easy when appropriate data has been inscribed in the graph. -> For every bacteria the amount of MIC needed has been captured and plotted in the graph with appropriate color coding for the antibiotics and the gran strain has been specified with the bacteria.

[malhotra-bhavini11]

The purpose of the visualization is to produce insightful and clear observations on the set of comparative data regarding antibiotic efficacy. There are 3 variables, 2 of which are categorical and 1 of which is numerical. The latter is the Minimum Inhibitory Concentration (MIC) which is the smallest concentration or volume required to halt bacteria growth. Gram staining is a test which indicates whether the bacteria colors blue (positive) or pink (negative) and is a quick way of separating the data into 2 classes.

With MIC on a logarithmic scale, we can observe through the data that the larger the concentration, the less effective an antibiotic is at inhibiting bacterial growth. We can also observe from the data that the MIC values have a huge range - from 0.001 to 870 so I thought to transform the values logarithmically. The effect of that is that the values are more evenly spread out and also reflect the log of the concentration as the time to grow the bacteria(antibiotic cultures grow exponentially) This visualization aims to answer 2 questions:

1. Does the bacteria group by resistance ?
2. Do different drugs correlate?

In order to answer the 1st first question, I have grouped bacteria into 2 groups: positive and negative gram staining within the visualization. This has been done by sorting the data so the values are grouped together and shaded in order to make a clear demarcation between the 2 groups. Another way this avenue could be explored is to group certain genus together (eg. Brucella abortus and Brucella anthracis ). I think that the table of data itself produced a clear image of the data except for the fact that any possible patterns are obscured by the sorting in alphabet order.

In order to answer the second question, I assumed a good choice for the visualization would be a heat map. In a heatmap, every number is represented by a switch of color. It could help in identifying extreme values and other patterns. In this case, the gradation of blue allows the viewer to quickly find extreme values without any difficulty. Additionally, the different antibiotics are the categories, so while it is easy to find extremes in the overall data, it is also easy to do so within one category by scanning down a column. From a horizontal perspective, it is easy to compare the MIC of the three different antibiotics for 1 particular pathogen.

As a statistics major, I believed that clustering the bacteria which react similarly to a particular antibiotic would be helpful from a research standpoint. Below is the dendrogram based on Fig 1. which was obtained through cluster analysis- it resembles Fig 1. However similar bacteria are grouped together which suggests further research in this area. Clustering can help build associations between data points and their features.

The graph thus shows:

• Penicillin is best for gram positive bacteria
• Neomycin is best for gram negative bacteria
• Streptomycin is best if the results of gram staining are unknown

Compared to other choices of representation such as stacked bar charts, it might be hard to distinguish the colors and it would be hard for viewers to overcome the ingrained mindset that the bars aren’t stacked but actually mean to show a shorter bar in front of a longer bar. There might have also been places where 2 bars coincide. This situation would decrease the readability of visualization since it would be difficult to differentiate the values of three antibiotics for 1 pathogen.

Meanwhile, using a grouped panel chart for all three antibiotics could cause sorting issues. This would cause confusion between sorting via efficacy for streptomycin or neomycin etc. given that the chart has 3 panels, each for one antibiotic. These 2 approaches discussed also do not cover the gram-staining characteristic of the bacteria that was present in the dataset.

[malhotra-bhavini11]

This is a representation of what manner the heatmap would appear like with the addition of clustering. This was the first time I did clustering analysis, so it is just acting as a demonstration of the type of graph as there is possibility for a mistake in this area.