Summary

The article discusses the challenging realities of data science jobs, which often differ from the expectations of aspiring professionals.

Abstract

Despite the high demand for data science roles, the article highlights that the actual job can be significantly different from the idealized version many learners imagine. Data scientists frequently encounter unexpected results due to data quality issues, lack of data, and inadequate problem formulation. The article suggests that data scientists must be adaptable and enjoy their work, similar to artists, even when faced with unpredictable outcomes. Additionally, the goals of data scientists, who prioritize data and research, often clash with business leaders focused on financial outcomes and project timelines. The piece emphasizes that addressing data quality issues is crucial, as it constitutes a significant portion of the job, contrary to the common belief that modeling is the primary task. The author also points out the health risks associated with prolonged screen time and the potential need to work with outdated technologies in some organizations.

Opinions

Data science jobs are not as straightforward or comfortable as they are often perceived, requiring dealing with unexpected challenges and results.
Data scientists should approach their work with an artist's mindset, finding joy in the process regardless of the outcome.
There is often a disconnect between the data-driven goals of data scientists and the financial objectives of business leaders, necessitating effective communication.
Addressing data quality issues is more critical than model optimization and should be prioritized.
The nature of data science work, involving long hours in front of screens, can lead to health issues.
Data scientists may need to work with outdated technologies in certain organizations, despite being knowledgeable in the latest advancements.

Your dream data science job may not be a bed of roses

In reality, it is far different than you imagine

Data scientist was the number 1 job role in 2020. However, in 2021, machine learning engineer is trending. There are still unfilled vacancies for data science professions in many countries.

Most people are studying data science and machine learning nowadays. Their ultimate goal will be getting a dream data science job. However, most of them don’t know the reality of a data science job as they are not dealing with real-world things while they’re learning the subject.

You may be familiar with different machine learning algorithms. You may also know the behind the scene process of each algorithm. You may also have the technical knowledge to implement each algorithm with data. Therefore, you may think the data science profession is a comfortable role. In reality, this is not the case. It is far different than you imagine. Let’s discuss some of the reasons for it.

Data scientists expect the unexpected

When it comes to the real world, this is the number 1 reason. Unlike programmers, data scientists may not get what they expect right away. Model building is also not a one-time task. They often need to go back and forth between the steps to build a better model. Even after doing that, they may not get what they expect. They expect the unexpected. Some possible reasons may be:

Data quality issue
Lack of data
Poor problem formulation without getting much domain knowledge

When data scientists get unexpected results, they should address the above issues. Of course, that is time-consuming. My opinion is that data scientists must think like an artist when they come to a solution. An artist enjoys what they’re doing whatever the result they get. Data scientists should enjoy like that. Sometimes, this will not be possible in real-world businesses where most of the things are measured in terms of money.

Business leaders measure most of the things in terms of money

Data scientists measure most of the things in terms of data. Therefore, the goals of the two parties may be totally different. Data scientists often want a significant amount of time to do research on the problem, do iterative tasks, improve the performance of the models, etc. This may exceed the project completion time. Business leaders often try to complete the project within the timeframe and get the payment. Communication between data scientists and business leaders will effectively eliminate these kinds of problems.

Data quality issues can break the entire project

Model optimization techniques can increase only a few percents of the model performance. But data quality issues can break your entire project. Such issues are:

Missing values
Outliers
Categorical features
High dimensionality
Multicollinearity
Lack of data

Addressing these issues should come first before doing any other tasks. Sometimes, your data may not capture the problem in your hand. It may be hard to find a good amount of data to solve your problem. Problem formulations are also tricky. Modelling is about 20% of the job and the rest will be the addressing of data quality issues and problem formulation. So, be prepared for that.

Data scientists spend hours of time in front of computer screens

This is the nature of the job. No one can change it. Spending hours of time in front of computer screens that emit harmful blue light will be the reason for many health problems. Your eyes and the brain may get strained. So, a data science job is not a comfortable job.

Some organizations use outdated technologies

Even if you’re up-to-date with the latest technologies for data science and machine learning tasks, you need to keep them aside and be familiar with outdated technologies that your organization use. Therefore, make sure to read the job description properly before you apply for the job.

Until the next time, happy learning for you! Meanwhile, you can read my other articles at:

https://rukshanpramoditha.medium.com

Rukshan Pramoditha, 2021–06–09