Giving an Overview of various current Job-Vacancies for University Graduates in Switzerland 🇨🇭
Main Purpose of the Analysis
Create a UI based on a Database.Where does the Data come from?
Email-NewsletterType of Data
TextNumber of Datasets used
1Number of Observations
20000Processing Techniques
Email-Scraping, NLP, JavaScript-DOMWhy should you care? 🤷🏻
The bigger Picture
Use Case 1
Automation of Job-Application Process.Use Case 2
Build a UI by having a Database in the Backend.Use Case 3
Use the Data to make Predictions.Current Openings for University Graduates? 🎯
The Results
FAQ
Programming the email-scraper, as well as using NLP-techniques for the relevant data extraction took around 1.5 days. Building the UI and deploying it to the cloud also took around one day each. In total, I estimate the effort for this MVP to be between 3.5 to 4 days.
- Using a cloud-connected database on the backend-side, which would communicate - in real time - with the UI.
- Giving a filter-option to the user. This would enable the website-visitor to aim for a more targeted search for specific job-positions.
- The jobs' associated images should be more targeted to match the advertised jobs (currently a randomizer takes over this role).
For cost & data security reasons, the database is synchronized offline (manually) with the UI. Furthermore, it is to note that only the most recent part of the database is published. Jobs which are very far in the past could theoretically also be listed on this UI, but this would be associated with the HTML being significantly bigger in terms of file-size. As a consequence, the site's loading time users would face would be (much) higher, resulting in a poor user experience.
The site's higher loading-time issues could be solved by choosing a so-called "server-side rendering" strategy - in combination with the "inifinite scrolling" technique, which could be implemented via the browser's "IntersectionObserver" API. Programmatically, such an application could be realized - for example - by using React. However, the condition to solve the loading-time issues requires the implementation of a real-time database on the server-side. Currently, however, pursuing a server-side rendering approach would not be worth it, because such an undertaking would incur too many costs.
Sure. The code is open source and can be accessed through my Github profile.