Note: This post is a guest contribution from the Department of Information Sciences & Technologies Rochester Institute of Technology.
The GIS field is emerging and creating new occupations across different industries. The United States Department of Labor has recognized GIS and broader geospatial technology as high growth industries with strong employment prospects. However, there currently is no model that can properly predict locations where GIS and geospatial technology occupations may become available. Thus, our research aims to develop a methodology for predicting GIS industry growth locations using ArcMap’s Model Builder. The result of our methodology will be validated for accuracy against the actual job offers to see if they are in locations predicted by our model.
Our project goals are two-fold. The first is to create a model which could predict geographic locations at the census tract level where job growth in the GIS field could occur. For our initial work, we have targeted New York and California. The second is to validate goal one results with geographic data of existing and past GIS job positions.
The base datasets that we used for our project are the 2010 Census Tracts provided by the United States Census Bureau for New York and California. By only using census tract-level data, we aimed to produce a fine-scale level of job predictability. The level of likelihood that a given location will have open positions in the GIS industry is summarized by a final score of a whole number value from 0 to 10, with 10 being the highest likelihood to have an open position, and 0 being the least likely to have an open position in the GIS industry.
Identifying growth in the GIS industry is difficult because the GIS industry as we define it today encompasses occupations in geology, cartography, and other fields which may be represented under different industry titles. To overcome this challenge, our team identified several indicators as contributors to GIS industry job growth. These indicators are educational attainment of the labor force, unemployment rate, previous job growth in technology industry, and proximity to transportation hubs. These indicators are processed by sub-models, which scale the individual indicators to a 0 to 10 range that then became part of the final score. Figure 1 shows an example of outputs produced by one of the sub-models.
Figure 2 shows an example of the final output for GIS job prediction in New York State. The green stars represent existing GIS jobs.
The final model of this project has received much useful feedback from professors in the Information Sciences and Technologies (IST) Department of Rochester Institute of Technology (RIT). From this feedback, our group recognizes that this model can be improved further to better predict the availability of GIS jobs within the United States. The first improvement is to include a variable that takes into account the proximity of universities in our model since firms usually establish their facilities near areas that have high number of colleges and/or universities. Second, we want to improve the model’s validation by including more actual GIS job offers from different job posting websites. Finally, we want our model to be able to predict not only at the census tract level but also from other level of administrative areas (for example, county level or block group level).
A video overview with more information about this work can be found at YouTube.
- Minh Quang Vo, Beau Bouchard, & Brian Tomaszewski, Ph.D.