A Data Science Story: Data Analysis in Education and the Digital Divide

By Tuly Reyes & Chelsea Prudencio

As promised in our series, “A Data Science Story” this blog will provide insights into data from surrounding villages in Nassau County as it relates to Education and the Digital Divide. The data set was downloaded from the US Census for the following Villages: Rockville Centre, Freeport Village, Garden City, The Village of Hempstead, and Lynbrook.

Step 1: In Data Science terms, we “wrangled” the data. That means, remove blanks, and organize in a structure that we can use. We used Excel, and a nice trick to “transpose” the rows and columns, then saved it as a comma separated values (CSV).

Step 2: Next we need to explore the data. So we will use Google Colab. It is an excellent tool for data exploration and analysis, again in Data Science terms, this is EDA or Exploratory Data Analysis.

During our EDA, we noticed that “computer and internet access” were reported in percentages. And while the percentages looked good all around, we wondered what the impact would look like not in percentages but in terms of individual persons.

Here is our analysis:

Step 1: Data Wrangling

Using the US Census data (estimates for 2019) we wrangled the data and created a Utility Matrix that we will use for the calculations. Since we are focused on the Digital Divide and how it affects education we used the following data fields:

  • Total House Holds
  • Persons Per House Holds
  • Percent of House Holds with Computers
  • Percent of House Holds with Broadband Internet Access

Step 2: EDA & Hypothesis

Lets look at the bar chart as percentages

As you can see there is no dramatics differences exposed in this visualizations. Our hypothesis, or question we asked ourselves was: Would the impact look the same if we converted from percentages to actual numbers?

So we compute the number of Households and the Number of Persons affected by the digital divide in each of these villages. We took the inverse of the percentages for computers and internet access and use them in our computation.

  • Total households without computers = Total Households x (1 – PCT With Computers)
  • Total households without internet = Total Households x (1 – PCT with Internet)
  • Total persons without computers = (Total households without computers) x (Person Per Household)
  • Total persons without internet = (Total Household without internet) x (Persons per Household)

The results are shown by the table below:

Step 3: Conclusion & Impact

By taking the percentages and converting them to numbers show the real impact the “Digital Divide” has on communities in our area. Sometimes showing impact a as percentage does not bring to light the seriousness of the problem. In the case of Hempstead we can see that the access to computers and internet affects 5,010 + 10,891. A total of over 15,000 persons are impacted by the digital divide.

Now let’s take a look at the visualization and not percentages but as actual persons affected.

The results are obvious. The digital divide impact is now clear between villages in our select data sets. As aspiring data scientist, we are anomaly spotters and we let the data speak for itself.