Olivia A. Gallucci

This is a step-by-step walkthrough of mapping U.S. cancer incidence and death rates in Tableau, using Python to check the data first. The short version: cancer death rates track poverty and income much more closely than incidence does. People in poorer areas don’t get cancer much more often, but more of them die from it.

This post is a report for MGIS 650, summarizing how to visualize cancer trends in Tableau for Rochester Regional Health.

Dear Rochester Regional Health (RRH), 

This guide summarizes how I created analytics and visualizations to explore cancer incidence in America and New York State. I hope that through this guide, RRH may easily duplicate and modify this work to fit its needs later. Please contact me if RRH would like any immediate changes. To complete this project, I used the provided sample datasets, but these steps can be followed with any dataset with the same structure as the sample dataset. 

My goals were as follows: 

1. Which regions of the country are most prone to cancer?
2. How do the factors affecting cancer vary by region?
3. Focusing on New York State, which factors are important? What should health professionals focus on to combat them? How should RRH proceed from a business perspective?

In each section, I will cover how I completed each task’s requirements, and the challenges I faced in the process.

Data Overview 

First, I explored the data to see if it needed to be cleaned. Here, I loaded RRH’s data to observe the structure of the data.

• Rows: 32,551
• Columns: 18

Initial Observations

RRH’s dataset includes cancer incidence and death rates, population estimates, poverty levels, and median income by ZIP code. This information will help analyze regional patterns and socioeconomic influences on cancer incidence. 

• incidenceRate and deathRate display the cancer incidences and effects.
• recentTrend, fiveYearTrend, and recTrend offer insights on cancer trends, which help identify regions with increasing or decreasing rates.

Data Quality Inspection 

Next, I checked for missing values, data types, and potential outliers using a basic Python program. 

	# before running this program:
	#
	# python3 -m venv myenv
	# source myenv/bin/activate
	# pip install pandas
	#
	import pandas as pd
	#
	# read file w encoding
	#
	file_path = '/Users/johndoe/Courses/mgis/cancerdeaths1.csv'
	data = pd.read_csv(file_path, encoding='ISO-8859-1')
	# display first few rows of dataframe to verify successful loading
	# data.head()
	#
	# check for missing values, data types, and
	# summary statistics for outlier detection
	#
	missing_values = data.isnull().sum()
	data_types = data.dtypes
	summary_stats = data.describe()
	print(missing_values)
	print(data_types)
	print(summary_stats)

view raw rrh-data-quality.py hosted with ❤ by GitHub

Here, I could determine that there are no missing values in the dataset, which is excellent. This meant I could proceed with the analysis directly without imputation or data cleaning.

Basic Statistical Analysis

Then, I examined distribution plots for some of the numeric columns (e.g., incidenceRate, deathRate, povertyPercent) to understand trends. Using Python, I generated visualizations for these columns.

	#
	# analysis part
	#
	import matplotlib.pyplot as plt
	#
	# list of numeric columns to examine for distribution
	# excluded: "PovertyEst", "popEst2015", "avgAnnCount", "avgDeathsPerYear"
	#
	numeric_columns = ["povertyPercent", "medIncome", "incidenceRate", "deathRate"]
	# create distribution plots for select numeric columns
	for column in numeric_columns:
	plt.figure(figsize=(8, 6))
	plt.hist(data[column], bins=30, edgecolor='k', alpha=0.7)
	plt.title(f'Distribution of {column}')
	plt.xlabel(column)
	plt.ylabel('Frequency')
	plt.show()

view raw rrh-basic-analysis.py hosted with ❤ by GitHub

The results of this program are as follows. 

Incidence Rate 

Downloadable

The distribution is roughly centered around the average, with a few areas showing notably higher rates.

Death Rate

Downloadable 

Similar to incidence rate, with most data clustered around the mean, but a tail of higher rates indicates some regions experience higher mortality.

Poverty Percent

Downloadable

This chart shows a skew toward lower poverty levels, but there are regions with significantly higher poverty rates.

Median Income

Downloadable 

The majority fall in the lower to mid-income range, with fewer high-income regions, indicating a skewed distribution.

Geographic Analysis

Outside of this basic analysis, I needed more sophisticated software to handle data manipulation, analysis, and visualization. To do this, I used Tableau. 

Tableau is known for its data visualization, and it is particularly effective for creating interactive dashboards and geographic maps (such as cancer incidence by ZIP code). It is also great for drag-and-drop analytics and interactive visualizations. 

In short, it will allow me to create clean, professional-looking dashboards with filters, hover-over details, and drill-downs—perfect for non-technical users. The downside of Tableau is that it is costly. To complete this project, I had to use a free trial of Tableau because I could not afford to pay for it.

General analysis: Which regions of the country are most prone to cancer?

This section seeks to answer objective 1: which regions of the country are most prone to cancer?

To do this, I will open Tableau, and import the dataset cancerdeaths1.csv. Once I successfully imported the data, I dragged State onto the map to create a geographical view. In Tableau, this view is known as a “sheet,” and I will be helping RRH explore the cancer data through these sheets. 

The dataset contains columns such as State, incidenceRate (rate of new cancer cases), deathRate (rate of cancer deaths), and population-related data (popEst2015). To identify regions in the USA that are most prone to cancer, I focused on the incidenceRate and deathRate across different states.

Sheets

I created two sheets: one focused on incidenceRate and another on deathRate. In each sheet, I placed incidenceRate and deathRate as color-coded measures on their respective maps. To do this, I dragged incidenceRate and deathRate over the color mark in each tab. Here, the color gradients represent higher rates with more intense colors.

Within both sheets, I used tooltips to make them more comprehensive. In Tableau, a tooltip is a pop-up box that appears when I hover over a data point, displaying additional information about that specific point based on the fields I add.

The tooltip information includes avgAnnCount, avgDeathsPerYear, and popEst2015 in the tooltip for additional context.

Lastly, I added filters to each sheet for State, incidenceRate, and deathRate to allow interactive exploration. In Tableau, filters allow me to refine and display specific data subsets in a visualization based on my criteria. The filters I selected enable RRH to explore the cancer data by narrowing down specific states, and adjusting ranges for cancer incidence and death rates, making it easier to identify regions with higher cancer prevalence and mortality for targeted analysis.

Overall, this setup will let RRH visually identify regions with the highest cancer incidence and death rates in the USA.

The charts above show the average cancer incidence and death rates by state. This view allows RRH to quickly identify states with higher incidence rates, which could be areas of interest for further investigation.

From these graphs, we can see that the states with the highest incident rates are New York, Pennsylvania, Texas, and California, while the states with the highest death rates are Texas, Pennsylvania, New York, and California. 

These graphs are located under the same headers in the Tableau workbook attached. 

Factor analysis: How do the factors affecting cancer vary by region?

By digging a bit deeper, I can also use this information to help form views of the second objective: how do the factors affecting cancer vary by region?

I already have maps of cancer incidence and death rates by state. Thus, I will duplicate these sheets, and explore potentially correlated factors like poverty percentage and median income with cancer incidence and cancer death rate. 

Through these sheets, you can see the poverty percentage and median income with cancer incidence and cancer death rate, and filter using the scales within each sheet. 

Map Insights

Poverty and income affect cancer outcomes, particularly in mortality. This could highlight socioeconomic barriers to healthcare access or differences in healthcare quality. 

On the other hand, cancer incidence has a weaker correlation with poverty and median income, suggesting that other factors may influence cancer occurrence more directly.

Scatter Plot Insights 

I also thought a scatter plot would be an interesting way to display this data. I began by creating a scatter plot for each factor to analyze potential relationships. To explore how cancer incidence may be related to external variables, I created separate scatter plots for each factor. I dragged cancer incidence to the Rows section and the chosen socioeconomic factor, such as poverty percentage or median income, to Columns. I made sure the Marks card was set to display data points, typically represented by circles. Then, I added a Trend Line by right-clicking on the plot and selecting “Trend Lines” > “Show Trend Lines” to gain a visual impression of the correlation.

The plot below shows states’ cancer incidence rates by their poverty percentages. 

Proof

Additionally, I performed a simple correlation analysis as proof using Python.

	import pandas as pd
	file_path = '/Users/johndoe/Courses/mgis/cancerdeaths1.csv'
	data = pd.read_csv(file_path, encoding='ISO-8859-1')
	# select columns for correlation analysis
	relevant_data = data[['incidenceRate', 'deathRate', 'povertyPercent', 'medIncome']]
	# calculate correlation matrix
	correlation_matrix = relevant_data.corr()
	print(correlation_matrix)

view raw rrh-correlation-analysis.py hosted with ❤ by GitHub

The correlation analysis reveals a moderate positive correlation (0.44) between cancer incidence and death rates, suggesting that regions with higher cancer incidence also tend to experience higher mortality rates. Similarly, a moderate positive correlation (0.44) exists between cancer death rates and poverty percentages, indicating that areas with higher poverty levels are more likely to have elevated cancer mortality. In contrast, there is a moderate negative correlation (-0.48) between cancer death rates and median income, implying that wealthier areas tend to experience lower cancer mortality. Additionally, as expected, a strong negative correlation (-0.77) is observed between poverty percentages and median income, with lower-income regions generally exhibiting higher poverty rates.

New York Analysis 

Important factors 

Then, I focused on the third question: Which factors are important in New York State?

I used ZIP code data to visualize incidence and death rates in New York. Here, I created a map visualization by dragging zipCode to Detail under the Marks card. Then, I placed incidenceRate and deathRate on Color or Size to visualize these rates by ZIP code. To filter for only ZIP codes in New York State, I dragged the State field to the Filters shelf, and selected “NY” to remove all other states. 

Then, I added filters to each sheet to display how median income, poverty percentage, incident rate, and death rate are distributed and impact each region. Here is a brief overview of the charts, noting that these charts do not show the exploratory power of the sheet on Tableau. 

Proof

I can also generate summaries for this data using Python. 

	import pandas as pd
	file_path = '/Users/johndoe/Courses/mgis/cancerdeaths1.csv'
	data = pd.read_csv(file_path, encoding='ISO-8859-1')
	# filter dataset for entries related to NY
	ny_data = data[data['State'] == 'NY']
	# calculate descriptive stats for required columns
	ny_analysis = {
	'Incidence Rate': {
	'Min': ny_data['incidenceRate'].min(),
	'Max': ny_data['incidenceRate'].max(),
	'Average': ny_data['incidenceRate'].mean()
	},
	'Death Rate': {
	'Min': ny_data['deathRate'].min(),
	'Max': ny_data['deathRate'].max(),
	'Average': ny_data['deathRate'].mean()
	},
	'Poverty Percent': {
	'Min': ny_data['povertyPercent'].min(),
	'Max': ny_data['povertyPercent'].max(),
	'Average': ny_data['povertyPercent'].mean()
	},
	'Median Income': {
	'Min': ny_data['medIncome'].min(),
	'Max': ny_data['medIncome'].max(),
	'Average': ny_data['medIncome'].mean()
	}
	}
	# data frame display
	ny_analysis_df = pd.DataFrame(ny_analysis)
	print(ny_analysis_df)

view raw rrh-ny-analysis.py hosted with ❤ by GitHub

The summary for New York data reveals that the incidence rate ranges from 435.2 to 577.4, with an average of approximately 497.5. The death rate spans from 138.8 to 198.4, averaging around 169, which indicates variability across ZIP codes. Socioeconomic factors show the poverty rate ranging from 6.3% to 31.5%, and median income varying widely between $33,687 and $98,312.

Dashboards 

The last step was building dashboards. A Tableau dashboard combines multiple visualizations, reports, and interactive elements on one screen, so charts, graphs, and maps sit together for easier exploration.

I built one dashboard for cancer incidence and mortality trends across the United States and another for New York State. Both put poverty rates and median income alongside cancer incidence and death rates, so RRH can see which regions have higher cancer prevalence and how socioeconomic factors influence cancer outcomes.

The New York section shows localized cancer trends, so RRH can prioritize intervention areas and allocate resources. For business stakeholders, it shows potential service gaps and areas where RRH can expand outreach or support. The dashboard is built for non-technical users.

What should health professionals focus on to combat factors affecting cancer incidence and death rate?

The incidence rate by state and ZIP code shows cases per unit population, which highlights regions with higher or lower occurrence. The death rate by state and ZIP code shows mortality. Paired with socioeconomic factors, it points to the populations most affected.

In practice, health professionals should focus on the ZIP codes where the death rate is high compared to the incidence rate. That gap usually means cancers are caught late or treated less, not that more people get cancer. The data points to three priorities: bring screening to high-poverty areas (for example, mobile mammography and at-home colorectal tests), remove access barriers such as transportation and insurance enrollment, and follow up on abnormal results quickly so patients start treatment sooner.

How should RRH proceed from a business perspective?

First, RRH should focus resources on at-risk ZIP codes with high incidence or death rates. Second, high-poverty or low-income areas with higher mortality are where preventive health campaigns and outreach programs should go. Third, the correlation between income and health outcomes can back funding applications or policy changes for underserved communities. Finally, tracking these metrics over time lets RRH respond to emerging public health challenges.

The poverty and income section has two metrics: poverty percentage, which may correlate with cancer incidence and mortality, and median income, which helps identify socioeconomic health disparities.

The demographics section shows average annual deaths, potentially segmented by location or demographic factors.

Lastly, filters let users narrow the data to a state (such as New York), an individual ZIP code, or a poverty or income bracket.

Final notes and challenges 

RRH can follow the steps above to replicate and customize the analysis as new data comes in. To help with Tableau’s learning curve, I am available for adjustments or questions as RRH begins applying this guide. The main takeaway: where people live affects whether they survive cancer more than whether they get it, so the biggest gains come from screening and access in high-poverty ZIP codes. If you liked this walkthrough, see how I used R and Tableau to track Dragon Ball’s brand sentiment on Reddit.