• When and How Atlantic Hurricanes Intensify: A Visual Analysis (1975–2024)

    –> A Visual Analysis of Atlantic Storm Activity (1975–2024) : Original dataset link: https://www.kaggle.com/datasets/utkarshx27/noaa-atlantic-hurricane-database?resource=download

    Objective & Purpose:

    Hurricanes are one of the most powerful and disruptive natural events, especially for coastal regions like Florida. Because of their impact on infrastructure, housing, and daily life, understanding when hurricanes occur and how intense they become is critical for preparedness and response.

    For this project, I selected the NOAA Atlantic Hurricane dataset (1975–2024) to explore patterns in storm frequency, timing, and intensity. The goal was to answer a few key questions:

    • When are hurricanes most frequent throughout the year?
    • Has hurricane activity changed over time?
    • What factors are associated with stronger storms?

    By analyzing this dataset, I aimed to uncover patterns that could help explain how hurricane behavior changes over time and what that might mean for future risk.

    Data Cleaning, Preparation, and Coding in R Studio

    Before creating any visualizations, the dataset needed to be cleaned and simplified.

    The original dataset included multiple entries for each storm, recorded every six hours. While this is useful for tracking storms in real time, it creates duplication when trying to analyze overall trends. To fix this, I summarized the dataset so that each storm appears only once, using its peak intensity values.

    Key Steps in R:

    • Loaded the dataset using read.csv()
    • Replaced invalid placeholder values (like -999) with NA
    • Filtered the dataset to focus on tropical storms and hurricanes
    • Created a proper date variable from year, month, and day
    • Grouped data by storm name and year
    • Calculated:
      • Maximum wind speed
      • Minimum pressure
      • Peak category
      • Primary month of occurrence

    This process allowed me to transform raw time-based data into a clean, storm-level dataset, making it much easier to analyze patterns.

    Visual Analysis & Key Insights:

    1. Hurricanes by Month (Ranking Analysis)

    This chart shows the number of storms by month.

    What it tells us:
    Hurricane activity is highly seasonal. The data clearly shows a spike in storms during late summer and early fall, with peak activity occurring between August and October.

    Question answered:
    When are hurricanes most frequent?

    2. Storm Activity Over Time (Time Series Analysis)

    This line graph tracks the number of storms per year from 1975 to 2024.

    What it tells us:
    There are fluctuations in storm activity year-to-year, but overall, the data suggests periods of increased storm frequency in more recent decades.

    Question answered:
    Has hurricane activity changed over time?

    3. Wind Speed Distribution (Distribution Analysis)

    This histogram displays how storm wind speeds are distributed.

    What it tells us:
    Most storms fall within lower to moderate wind ranges, while high-intensity storms are less frequent but still significant. This highlights how extreme storms are rarer but impactful.

    Question answered:
    How are storm intensities distributed?

    4. Wind Speed vs. Pressure (Correlation Analysis)

    This scatterplot compares maximum wind speed with minimum pressure.

    What it tells us:
    There is a clear inverse relationship, storms with higher wind speeds tend to have lower pressure. This aligns with meteorological principles, where stronger storms are associated with lower central pressure.

    Question answered:
    What factors are associated with stronger storms?

    Conclusion & Reflection:

    This project demonstrated how visual analytics can turn complex datasets into clear, meaningful insights. By cleaning and summarizing the data, I was able to focus on the bigger picture rather than getting lost in repeated time-based observations.

    One of the most interesting takeaways was how predictable hurricane seasonality is, with a clear concentration in late summer months. At the same time, variability in yearly activity suggests that other environmental factors likely influence storm frequency and intensity.

    Overall, this analysis highlights how data can be used not just to observe patterns, but to support better planning and awareness, especially in regions where hurricanes are a recurring concern. Understanding when and how hurricanes form isn’t just about data, it’s about being better prepared for what’s ahead.

  • Assignment Objective:

    “Build and visualize a simple un-directed social network graph using Python. You’ll use nodes and edges to represent people and their connections.”

    Python Code:

    Walkthrough of Code:

    For this assignment, I used Python along with the libraries NetworkX, pandas, and plotnine to create a social network visualization.

    First, I imported the necessary libraries. NetworkX was used to create and manage the network graph, pandas helped organize the data into data frames, and plotnine was used to visualize the graph in a ggplot-style format.

    Next, I created a random undirected graph using nx.gnp_random_graph(10, 0.5). This generated 10 nodes with a 50% probability of connections between them. I then assigned labels A–J to each node so the graph would be easier to read.

    After that, I used a spring layout (nx.spring_layout) to position the nodes. This layout spreads the nodes out in a visually balanced way based on their connections.

    I then converted the graph into two data frames:

    • nodes_df, which stores the x and y positions of each node along with its label
    • edges_df, which stores the starting and ending coordinates for each connection (edge)

    These data frames were necessary because plotnine requires structured data to create visualizations.

    Finally, I built the visualization using plotnine:

    • geom_segment() was used to draw the edges (lines between nodes)
    • geom_point() was used to plot the nodes
    • geom_text() added labels (A–J) above each node
    • theme_void() removed unnecessary background elements
    • ggtitle() added a title to the graph

    Issue Encountered (and how I fixed it)

    One issue I ran into was that the graph would not display when running the code in Python IDLE. Instead of showing the visualization, the output only displayed:

    <ggplot: (640 x 480)>

    This means the plot was successfully created, but Python IDLE did not automatically render or display it.

    To resolve this, I modified the code to save the graph as a PNG file instead of trying to display it directly. I used:

    network_plot.save("network_graph.png", width=6, height=4, dpi=300)

    This allowed the visualization to be saved as an image file in my Downloads folder. I was then able to open the image and use it for submission.

    Network Graph:

    Reflection:

    For this assignment, I created a social network graph in Python using NetworkX, pandas, and plotnine. What worked well was generating the random network and converting the node and edge data into a visualization with labeled connections. One issue I encountered was getting the graph to display in Python IDLE, but I resolved it by saving the graph as a PNG file instead. This method worked well overall because it allowed me to clearly visualize relationships between nodes in a network. While I initially had trouble displaying the plot, saving it as an image provided a simple and effective workaround. I would use this method again for basic network visualization tasks because it is flexible and produces clean, readable outputs. Overall, I would use this method again because it is a simple and effective way to visualize relationships between nodes in a network.

  • Assignment Instructions:

    Using the data given from the module instructions, data was constructed in R:

    Next, it was time to make the visualization:

    Visualization Created:

    A dot-dash style plot recreated in ggplot2 to illustrate a Tufte-inspired approach to visual analytics. The chart highlights changes across time while using dashed reference lines and direct labeling to improve clarity and comparison.

    Explanation of the visualization

    For Module 11, I chose to recreate a dot-dash plot in ggplot2 based on the work discussed in Dr. Piwek’s post on Edward Tufte and Charles Minard. This visualization shows a time-based trend using points and a connected line, while dashed horizontal reference lines make important value levels easier to compare. The design reflects Tufte’s emphasis on clarity, minimal clutter, and meaningful comparison. Instead of relying on decorative elements, the chart focuses attention on the underlying data and uses labels directly in the figure to support interpretation.

    Connection to Tufte:

    This plot connects to Tufte’s ideas because it presents information in a simple and efficient way. The graph avoids unnecessary colors and distracting design choices, which helps the viewer focus on the pattern in the data. Like the Minard visualization discussed in the notes, it also integrates text and graphics in the same space rather than separating explanation from evidence. Overall, the chart demonstrates how visual analytics can make trends and comparisons easier to understand.

  • R Code Setup:

    I began by loading the ggplot2 package and accessing the built-in economics dataset, which contains time-based economic indicators.

    Creating the Time Series Visualization:

    Next, I created a time series line graph by mapping the date variable to the x-axis and median unemployment duration (uempmed) to the y-axis.

    Visualization Output:

    Discussion:

    For this assignment, I used the built-in economics dataset in R to create a time series visualization using ggplot2. I chose to analyze the median duration of unemployment (uempmed) over time. The line graph shows how the number of weeks individuals remain unemployed has changed across different years.

    From the visualization, I observed that the duration of unemployment fluctuates over time, with noticeable increases during certain periods. These peaks likely reflect times of economic difficulty, where individuals remain unemployed for longer periods. This demonstrates how time series visualization helps reveal trends and patterns that are not immediately obvious from raw data.

    In my R code, I mapped date to the x-axis to represent time and uempmed to the y-axis to show changes in unemployment duration. The geom_line() function was used to create the continuous trend line, while labs() and theme_minimal() helped keep the graph clear and easy to interpret.

    This assignment also connects to the work of Charles Minard, whose visualization of Napoleon’s 1812 campaign combined multiple variables to tell a powerful story over time. While my graph is much simpler, it still shows how time-based data can be used to communicate meaningful patterns and trends.

  • Assignment Details & Objectives:

    Using ggplot2 in R studio, I used the following code to then create the following visualization:

    Write-Up Questions & Discussion:

    For this assignment, I used the mtcars dataset to explore multivariate relationships using a scatter plot created in ggplot2. I chose the built-in mtcars dataset because it contains multiple numerical and categorical variables that are ideal for multivariate analysis. It includes features like weight, horsepower, displacement, and cylinder count, which are all related and allow for meaningful comparisons in a single visualization. I visualized the relationship between weight and miles per gallon while incorporating additional variables through color and size. Specifically, cylinder count was represented by color and displacement by size, allowing multiple variables to be analyzed within a single visualization.

    From the graph, I observed a strong negative relationship between weight and mpg, where heavier cars consistently have lower fuel efficiency. Additionally, cars with more cylinders (shown in blue) tend to cluster toward higher weights and lower mpg values, while 4-cylinder cars are generally lighter and more fuel efficient. The size encoding also shows that cars with larger displacement tend to appear in the lower mpg range, reinforcing the idea that multiple engine-related variables are interconnected.

    Using color and size as visual encodings made it possible to represent several variables without overcrowding the plot. This aligns with Few’s design principles, particularly the emphasis on clarity and maintaining a strong visual correspondence between data and representation. The use of a simple color scheme, consistent scaling, and minimal background (theme_minimal) helped keep the focus on the data rather than unnecessary visual elements. However, this assignment also showed that multivariate visualizations can become complex quickly, so it is important to balance detail with readability.

    Overall, this visualization made it easier to identify relationships between multiple variables at once, highlighting patterns that would not be as clear in a simple two-variable analysis.

  • R Studio Instructions:

    R Studio Code:

    Data Visualization:

    Module Analysis:

    For this assignment, I used the built-in mtcars dataset to explore relationships between miles per gallon (mpg) and several vehicle attributes, including displacement (disp), horsepower (hp), and weight (wt). After creating scatter plots with regression lines in ggplot2, I observed mostly negative relationships between mpg and all three variables. In particular, weight showed one of the strongest patterns, where vehicles with higher weight tend to have noticeably lower fuel efficiency. Horsepower and displacement showed similar downward trends, suggesting that as engine size and power increase, mpg generally decreases.

    Using a facet grid layout made it much easier to compare these relationships side by side. Instead of overlaying all variables in a single plot, separating them into individual panels allowed the trends to be interpreted more clearly while still having the same visual structure. This aligns with the idea discussed in lecture that grid layouts improve comparison and reduce visual confusion.

    Few’s design recommendations were also helpful for keeping the visualization clear and readable. By using a minimal theme, simple color choices, and clear axis labels, the focus stays on the relationships in the data rather than decorative elements. I think Few’s approach generally improves visual analytics because it prioritizes clarity and interpretation, although sometimes extremely minimal design can feel a bit plain. In this case, however, the simplicity helped emphasize the regression patterns.

    Additionally, my DataCamp experience from my Data Mining class helped me feel more comfortable working with ggplot2 and interpreting relationships between variables. Having practiced building scatter plots and regression models previously made it easier to structure this visualization and understand what the regression lines were showing.

  • Assignment Instructions:

    Coding in R:

    First it is important to load the R packages necessary for the assignment, in this case, ggplot2. Also, I am going to load the data set in R “mtcars” for this visualization exercise.

    Next, I converted the variable “cyl” from a numerical variable to a factor. This is important for creating my Histogram chart. Converting “cyl” to a factor tells R to treat each cylinder type as a distinct category, rather than a numeric scale.

    Lasty, I used ggplot2 to create a faceted (side by side) histogram. Using density rescales the histogram from raw counts to probability density, which allows the smooth density curve to align properly and emphasizes the overall shape of the distribution rather than frequency alone. The facet_wrap() function creates separate but aligned panels for each cylinder category, making side-by-side comparison clearer and more consistent with Few’s recommendation for comparable scales. theme_minimal() reduces unnecessary visual clutter so the focus remains on the data itself rather than decorative elements.

    Histogram:

    Reflection Paragraph:

    For this assignment, I used the built-in mtcars dataset to examine how miles per gallon (mpg) is distributed across vehicles with different cylinder counts. The faceted histogram with a density curve shows that 4-cylinder cars tend to cluster at higher mpg values, while 8-cylinder cars are concentrated at lower mpg levels. The 6-cylinder distribution appears more tightly grouped in the mid-range. The density curves help emphasize the overall shape of each distribution, making differences in spread and central tendency easier to see.

    In terms of design, I followed Few’s recommendations by aligning the axes across all panels and avoiding unnecessary visual clutter. The faceted layout allows for side-by-side comparison without overlapping distributions, which improves clarity. The consistent bin width and minimal color usage also support accurate comparison. I agree with Few’s critique that distributions are often presented in ways that make comparison difficult; in this case, keeping the scales consistent and the design simple made the differences much more clear and easier to interpret.

  • Assignment Instructions:

    Creating a Pie Chart in R:

    Creating a Bar Chart in R:

    Creating a Histogram in R:

    Creating a Boxplot in R:

    Discussion:

    These basic visualization graphs in R helps show simple patterns in the data, especially with the histogram and boxplots using the mtcars dataset. The histogram of MPG allows us to see the overall distribution of miles per gallon and identify where most vehicles fall, while the boxplots make it easier to compare MPG across different cylinder groups.

    In terms of simple descriptive analysis, these visuals do fit with the ideas discussed by Few and Yau because they allow us to quickly see differences, variation, and possible outliers without overcomplicating the design. For example, the boxplot clearly shows that cars with fewer cylinders generally have higher MPG, which highlights a meaningful pattern in the data. However, these are still very basic graphics and do not incorporate more advanced techniques like deviation analysis or multi-variable comparisons. Overall, the visualization works well for introductory descriptive analysis because it communicates patterns clearly and efficiently, even if it is not as detailed or analytical as more advanced dashboard designs.

  • Ranking Chart:

    Part-To-Whole Chart

    Why I Chose This Dataset:

    I chose this dataset because it provides a simple but effective way to explore how values change over time. Even though it only includes two variables, Average Position and Time, it allows for meaningful analysis through different visualization frameworks. It is interesting to see how a relatively small dataset could still reveal patterns when structured intentionally.

    The Story and Insights the Visualizations Tell:

    The Ranking chart clearly shows a steady increase in Average Position as time progresses. By ordering the values from highest to lowest, the upward trend becomes immediately visible. The later time intervals dominate the top of the ranking, indicating consistent growth rather than random fluctuation.

    The Part-to-Whole donut chart reinforces this pattern from a proportional perspective. When time is grouped into ranges (0–1, 1–2, 2–3, 3–4), the later ranges account for a significantly larger share of the total Average Position. This confirms that the higher values are concentrated in the later stages rather than evenly distributed across time. Together, the two visualizations complement one another: Ranking highlights relative order, while Part-to-Whole highlights proportional dominance.

    Reflection: Strengths and Limitations of Part-to-Whole:

    Part-to-Whole visualizations are strong tools for showing distribution and concentration within a dataset. They quickly reveal which segments dominate and how contributions compare proportionally. In this case, the donut chart effectively demonstrates that the 3–4 time range contributes the largest share to the total.

    However, Part-to-Whole designs also have limitations. They can become cluttered if too many segments are included, and precise comparisons between slices can be difficult. Without grouping the data into ranges, the chart would have been visually overwhelming and harder to interpret. This highlights the importance of thoughtful data preparation before visualization.

  • Assignment Instructions:

    Created graphic:

    Tableau Link: https://public.tableau.com/app/profile/sydney.claudio/viz/Claudio_Transit_Ridership_TimeSeries_Module4/Sheet1

    Explanation/Visual Analysis:

    For this visualization, I selected six variables that helped create a clear time-based comparison of public transit usage across different U.S. cities. I focused on variables that captured both the time component (Year), geographic context (Primary USA City), and operational performance (Ridership and related transit service metrics). Reducing the dataset allowed me to focus on the most meaningful indicators while keeping the visualization readable and easy to interpret.

    The visualization reveals a few important patterns. First, there is a general downward trend in ridership across most selected cities over the time period shown. Los Angeles shows the highest ridership levels overall but also demonstrates a noticeable decline toward the most recent year. Miami and Atlanta show smaller but similar directional patterns, suggesting broader influences affecting transit usage rather than isolated local changes. St. Louis maintains consistently lower ridership levels, indicating differences in transit reliance between cities.

    Overall, the time series line chart effectively highlights long-term trends, allows for city-to-city comparison, and supports pattern recognition through temporal visualization. This type of visualization is especially useful for identifying system-wide changes, potential external influences on ridership, and differences in urban transportation demand over time.

Create your own website, free on WordPress.com.Create your own website.
Start free