Employees spend a large portion of their lives at work each day. Providing a healthy and sustainable work environment helps to promote productivity and create a culture of employees that value the health of their surroundings. To sustain a safe, healthy and functional workplace, an organization must train its employees the practical and efficient work processes to minimize the impact of employee production on the office environment and employees. Sustainable workplace practices go beyond what is required by law and ensure longevity and overall well-being of the workforce.

What it does

The company has been conducting several sustainability workshops to promote a sense of responsibility towards the environment in its employees. How often do you, as an employee, involve yourself in sustainability practices? Quite often? Then, be interested in we finding out if you will continue to practice sustainability methods in your future.

How we built it

We have built it as an Artificial Neural Network, using Machine Learning.

Challenges we ran into

Data preprocessing was the most difficult part! All the variables available to us can be divided into independent variables – Index, RowNumber, EmployeeID, Surname, TrainingScore, Geography, Gender, Age, TrainingLevel, HasRewardCard, IsActiveMember and the dependent variable – Exited. Based on the independent variables, we will be predicting the outcome of the dependent variable. In order to build an efficient model, we need to consider our variables with care. Since RowNumber, has no impact on whether the employee exits the training or not, it will not be included in our ‘matrix of features’. The same goes for EmployeeID and Surname. The variables that we consider in our ‘matrix of features’ will be TrainingScore, Geography, Gender, Age, TrainingLevel, HasRewardCard, IsActiveMember. Once the variables are selected, we need to check for categorical variables among them so as to carry out the process of encoding. We have independent variables – Geography and Gender, that have categories as strings. Hence, these variables must be encoded before being used. Also, since Geography has three values associated with it, we need to create dummy variables for this variable, and even make sure to avoid the dummy variable trap.

Accomplishments that we're proud of

Our model boasts an incredible accuracy of 83.35%!

What we learned

Feature Detection and Selection, Creating deep learners to make learning more human-like.

What's next for Sustainable Future?

Develop a group of models that could predict environmental issues such as deforestation, air quality and so on based on different datasets. With the growing importance of data today, it is not wrong to say that data will form the basis of environmental education in the near future.

Built With

Share this project: