ArcGIS Insights & Python Script Project

The data used as a base of the Covid-19 projections separated British Columbia into five different regions: “Fraser,” “Interior,” “Vancouver Island,” “Northern,” and “Vancouver Coastal,” (see figure 1.1), and were stored under the classification of “countyname.” The Regional Health Authority dataset collected data from those regions can be seen in table 1.1. More specifically, they cross referenced the Canadian regions with their populations, total cases, hospitalizations and dates of infection.

The data clearly shows that on average, there is an increase in Covid-19 cases across all regions. Between March 23rd and March 30th, Fraser, the region with the largest population, had a 98.16% increase in total cases. Similarly, the hospitalizations in the same region more than doubled, increasing by 122.22%. These numbers are consistent even among regions with the smallest population, such as Northern, whose total Covid-19 cases increased by 133.33%.

Figure 1.1 Map of territories in British Columbia, used to chategorize data regionally.

After analyzing the initial input data further, I scripted the three CHIME prediction models into ArcGIS Insight’s coding module. In Figure 2.1, we can see the code used to script these models, by first defining the SIR variables, and running projection functions through the existing data to create the predictions.

The CHIME model uses its own global pandemic data parameters, to cross reference with the input data to make current predictions. More specifically, the results focus on the infection trends between September to December 2023, in the same British Columbia Regions. The model then groups the cases into the SIR categories: “Susceptible,” “Infected,” and “Recovered.” In order to provide hospitals with further useful information, the “infected” data is then sub categorized into “Hospitalized,” “ICU,” and “Ventilated” variables, which are overlapping. In Table 2.1, we can see that the average number of people who are predicted to be infected between September and December of this year reaches the thousands in every region. For example, the average number of infected people in the region with the largest population, Frasier, is 186,137.11, which is 10.82% of its 2020 population. Moreover, the predicted number of susceptible citizens of the same region consists of 53.32% of their population. While these numbers are seemingly intimidating, the predicted number of recoveries is far higher than any other category, reaching 616,723.72 in Fraser.

Figure 2.1 Portion of Python code used to script in ArcGIS Insights, displaying the initial prediction model codeing process

Figure 2.2 Pie chart map of infection status, highlighting Fraser's predicted infection count

Figure 2.3 Pie chart map of infection status, highlighting Northern's predicted infection count

The final, most useful, visualization displays the average predicted values of the same four “status” categories, over the next four months of 2023. As seen in figure 2.4, the average number of susceptible people across all regions is significantly higher than any other category, reaching a value of over a million. At the same time, the infection, hospitalization and recovery estimations are very low. What this tells us is that August and September are prime months for Covid-19 incubation. More specifically, as Covid follows common viral patterns of infection waves throughout the year, high susceptibility levels imply the beginning of a new wave.

This is corroborated by the data in figure 2.5, where susceptibility is predicted to plummet, converging with rising rates of infection and recovery. This crossover implies that we will see the most active effects of Covid-19 in early November. While people will be generally less susceptible to the virus, we will also see an influx in the virus’s activity. As we can see, hospitalizations rise significantly less than the other two categories. This proves that while the virus is still prevalent in our daily lives, it is far less fatal than it used to be at its inception. Finally, as shown in figure 2.6, by December 2023, it is predicted that the susceptibility and infection levels will decrease significantly, while recovery rates will rise. Within the last month of 2023, the number of infections are anticipated to have dropped by 87.29% on average. This clearly marks the end of this year’s Covid-19 viral wave.

Calculating the predicted viral trends of Covid-19 might not seem relevant anymore because of its decrease in severity. Nevertheless, this type of data is critical for anticipating the resources and alternative preparation necessary for dealing with such an illness. Just like the Flu, Covid continues to mutate and return in waves throughout the year. By creating prediction models for its patterns, we can become more accustomed to its trends, thereby attempting to manage the viral spread and potential fatalities that could result.


















Watch my full process of scripting and creating the basic data visualizations in ArcGIS Insights:

Figure 2.4 Average value of projected Covid-19 status over time, highlighting September case numbers at suseptability peak

Figure 2.5Average value of projected Covid-19 status over time, highlighting November case numbers at status crossover

Figure 2.6Average value of projected Covid-19 status over time, highlighting December case numbers at infection low