Job Risks in the USA - Occupation Vs Automation
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Introduction
Throughout history, technological progress has vastly shifted the composition of employment - from manufacturing and clerking to service and management occupations. In this era of Automation, artificial intelligence, and robotics have created the potential to penetrate deeply into occupations with steadily decreasing labor cost. As businesses across industries have started the transition to digital operations, many job positions are being deemed redundant.
This project is to highlight those risk area so that proper audience can plan ahead to reduce unemployment in different states of US. The main audience for this project is the HR departments of all companies along with Government workforce teams in different states. Companies should report planned workforce changes to a global automation observatory as this gives workers time to look for a job before becoming unemployed based on the prediction status and governments/civil society time to plan for specific reskilling, tax revenue forecasting, and recruitment of new businesses.
Dataset
The dataset is retrieved from Kaggle dataset base – this dataset is made by Carl Benedict Frey and Michael Osborne who did research work on the impact of computerization with the primary objective of analyzing the number of jobs at risk and to find out the relationship between an occupation’s probability of computerization, wages and educational attainment. The dataset consists of 702 different job titles and their probabilities to automation.
Application
An interactive shiny web application has been developed using R, ggplot2, gvismap and shiny dashboard to analyze the current status of the job positions and automation risk probabilities. The web app screen consists of 5 different links to explore the current status of job automation. Following are the detail functionalities and visualization of each tab.
Overview of Automation Job Risks
This screen gives an overall statistic of the most impacted and least impacted states with risk counts of job positions. State based job counts is an interactive map hovering over which, the counts of job loss is highlighted based on automation probability.
Red highlighted states like Texas, California are already in high risk as the number of manual job counts are high in those places which can be automated in near future completely.
This screen contains another histogram, which shows the count of job titles among the different ranges of high, medium and low probability levels.
Explore by Job Title
This link provides the functionality to search by job title and get a statistic of different features. People can find out what are the top safe states for that job position through this functionality and can take decision to get relocated to safe states. Additionally, user can check how many jobs are there in every state for that job title. A very user-friendly interactive functionality has been provided in the screen to look for automation probability of similar kind of title for comparison.
As an example, if user wants to search the status of job title as 'Advertising and Promotional Managers', the top safe states for this title looks like following:
The gvismap will show count of jobs with the same title as 'Advertising and Promotional Managers' for every US state.
If user wants to check the probability of automation for other similar manager title, the bar chart shows the exact statistics.
Explore By State
This link provides the functionality to check the automation status by state name. User gets the interactive visual bar chart for risky jobs in that state and scatter chart for overall job density on that state.
As an example, if someone wants to check the impact of automation for the state of Massachusetts, following is the graph of top risky jobs.
Overall job density of Massachusetts shows that almost 25% of the job in this state are with more than 75% automation probability.
An additional user-friendly interactive feature has been provided here to check the count of employees impacted with similar job title in this state. For an example, if we want to apply this feature for 'Clerk' job title, below is the bar chart showing almost 60000 office clerks are in high risk. More than 45000 Stock clerk and Order Fillers are impacted as well. This feature provided a easy comparison to get overall idea on the risk of that job title.
Summary of Job Risks from Automations
This link gives a full summary of the data analysis done through visualization and interactive operations and represents four different graphs on concluding the automation risk on Job title.
All over US, here are the top 10 Job Title which will impact majority of people in next decades. It is observed that Retail Salesperson and Cashiers are in vulnerable position.
Job Category plot shows that Sales and Clerical Departments are the two high risk job category whereas Education is still in safer zone.
Automation probability density scatter plot shows that the density is very high for the job titles where automation probability is more than 85%. That is extremely concerning.
The final summary is a column chart which gives the statistics of most impacted top 10 states with the employee counts due to job automation. California, Texas and New York are in first three where millions of people can lose their job if not re-skilled before the crisis time approaches.
Based on above mentioned analysis, the final conclusion comes as 49% of total US jobs are in high risk which is a high number. These statistics will help the HR team of different organizations, Government work force team and the workers for understanding the emergency and plan ahead for reskilling or upskilling to avoid the danger of job loss for millions of people.
For R code, please click here
About Author
