HR-Salifort-Motors

Predictive Models Salifort Motor

In this project, we are creating a predictive model for a fictional Motor company ‘Salifort Motors’ to ancipate whether or not an employee in said company is likely to quit. This is to figure out the factors that contribute to people quiting and what the company needs to focus on to minimize turnover. We will implement feature engineering, model development, and evaluation to see what factors the HR deparment at Salifort Motors should point their attention to.

Background on the Automatidata scenario

The HR department at Salifort Motors wants to take some initiatives to improve employee satisfaction levels at the company and minimize turnover. They collected data from employees, but now they don’t know what to do with it. They refer to us as a data analytics professional and asks us to provide data-driven suggestions based on our understanding of the data. They have the following question: what’s likely to make the employee leave the company?

The goals in this project is to analyze the data collected by the HR department and to build a model that predicts whether or not an employee will leave the company.

If we can predict employees likely to quit, it might be possible to identify factors that contribute to their leaving. Because it is time-consuming and expensive to find, interview, and hire new employees, increasing employee retention will be beneficial to the company.

Assignment

In this project, we will complete the following deliverables:

Data Clean and Organize

Analyze important factors

Design and implement a machine learning model

Re-analyze important factors

Draft an executive summary of your results

Data Cleaning and Organizing

Given the size of the data, and the research we were doing, we felt the dataframe provided had sufficient data to derive valuable information from. Therefore, in the planning part of PACE, all we focused on were missing data, outliers, and duplicates to decide what actions to take against them, as seen in the Data Cleaning Code.

Data Analysis

In data analysis, we first decided to analyze the relationship with numbers of hours worked and number of projects worked in relation to people who quit. This is because we felt, realistically, these two factors would/could be very big clues on the quality of life for the employees and could anticipate which segment of people would leave, and which wouldn’t.

As seen in the data, those who worked more hours as well has more project had higher likelihood of quitting. For example, with those working 7 projects and consequently more hours, every single one of them had quit. Looking one step lower, about 50% of those working 6 projects had quit. With more hours and more projects, came higher likelihood of employees quiting. There are outliers, however, with those working the lowest amount of projects and hours having high exit rate as well. This data could more so be due to quite quitter who were soon leaving the company and deciding to have less work or those not doing much work and being fired than it does have to do with people actually working those hours and deciding to exit themselves. This would be an information that we would pointed out to Salifort Motors and ask for more information on to dive deeper into said subject.

We also ran a plot on the relationship between leavers, satisfaction level, and the amount of hours worked a month.

This plot showed that there were masses of those working over 240 hours a month (about 75 hours a week) with very low satisfaction rate as well as those working normal hours with low satisfaction rate. While the reason for low satisfaction rate with the groups working over 240 hours a month is self-explanatory(worse work-life balance), those working normal hours with low satisfaction rate is a little bit trickier. As explained before, this could be those who decided to work less hours as they were already exiting the company who got mixed into the data of low hour workers. It could, however, also be those who didn’t like the lifestyle of working less than their peers and feeling the pressure to work more from their peers and managers, leading to lower satisfaction. Yes, there are those with high hours and high satisfaction rate who did leave, but that could simply be explained by better job offers, or other personal reasons.

We also made a visualization on how tenure affected satisfaction rate.

This showed a higher average of satisfaction rate with employees with over 6 years tenure, and those under 3 years tenure. Employees that presented lower satisfaction rate and were more likely to leave the company were those with mid numbered tenure years from 3 to 6 years tenure. This is very interesting data in regards to who to focus on when trying to minimize turnover rate.

We then compared salary, which showed that the long-tenured employees group comprised of more high-paid employees than the other groups.

Comparing evaluation scores and hours worked with those who left vs. stayed, with a scatterplot, we figured employeed who worked the most hours and had very high evaluation scores were far more likely to quit. We also saw those who worked less hours with lower evaluation scores left. This second observation could be due to these individuals getting fired more than them quitting, and the first observation could be due to very intense work-life balance. We also noticed that having more hours doesn’t give you a better evaluation, as the evaluation rating was spread across the board in regards to the various amount of hours worked.

We then plotted the relationship with amount of hours worked, and whether they got a promotion or not. We came to find that very few employees working extensive hours were promoted. All of the employees who left were the ones working the longest hours without getting a promotion. This creates a very clear image of why those with mid ranged tenures with the most hours are far more likely to leave than those just starting out or those who had been there over 6 years. Regardless, whether working a lot of hours or not, employees noticing the amount of work they put in has nothing to do with promotion does prompt people to consider leaving especially after staying multiple years and actually seeing no reward for their commitment career-wise.

We also plotted to see whether the departments employees worked in had any effect on whether they’d quit or not. It did not.

Creating a heatmap

Creating a heatmap, we found that that the number of projects, monthly hours, and evaluation scores all have some positive correlation with each other, and whether an employee leaves is negatively correlated with their satisfaction level. Meaning higher satisfaction results in lower chance of leaving.

Insights

It appears that employees are leaving the company as a result of poor management. Leaving is tied to longer working hours, many projects, and generally lower satisfaction levels. It can be ungratifying to work long hours and not receive promotions or good evaluation scores. There’s a sizeable group of employees at this company who are probably burned out. It also appears that if an employee has spent more than six years at the company, they tend not to leave; more so due to loyalty.

Design and implement a machine learning model

Upon testing Logistic Regression model with an F1 score of .80, we also tested a Random Forest Classifier whose F1 score came out to .87 after training. We decided to use the Random Forest Classifier as our learning model as it has a better possibly of prediction. Our codes could be found in the Random Forest Code page.

With implementing all variables as features, we created a Random Forest Classifier that was trained and has the ability to predict whether an employee will quit with an F1 score of .87/87% success rate.

Creating a Confusion Matrix on the Random Forest Classifier, we are able to see that the model predicts more false positives than false negatives, which means that some employees may be identified as at risk of quitting or getting fired, when that’s actually not the case. Regardless of that, this is still a strong model with a 87% F1 score.

From the Random Forest Classifier, we were also able to figure out which columns/factors were the most important in predicting whether an employee would quit or not. This information as well as a Random Forest Classifier of 87% gives the Salifort Motors team very powerful tools on what to look for in minimizing turnover, and more specifically, who to look for and pay attention to.

Summary of Analysis and model results

After conducting feature engineering, the random forest modestly outperformed Logistic Regression with an 87% F1 Score compared to 80% for the Logistic Regression.

With the Random Forest Classifier and previous analysis, we were able to see that evaluation score, number or projects, tenure, and amount of hours employees worked had the biggest impact on employees leaving or staying.

While realizing that putting in more work had no benefit and didn’t necessarily mean being more likely to be promoted, those working the most hours and the most projects in the company are far more likely to depart from the company than any other group. Taking into account that they aren’t new to the company, and understand this information, nor loyal to the company and are willing to depart, those with a mid range tenure are far more likely to depart the company; especially those not receiving promotions within 5 years. In addition, those with high evaluation scores and working at the extreme hours are those with far higher chance of leaving the company, meaning the likelihood of retaining their best and hardest working performance is worst than retaining any other group, which is really bad for the company’s longevity.

Executive Summary

The models and the feature importances extracted from the models confirm that employees at the company are overworked.

To retain employees, the following recommendations could be presented to the stakeholders:

Cap the number of projects that employees can work on. Consider promoting employees who have been with the company for atleast four years, or conduct further investigation about why four-year tenured employees are so dissatisfied. Either reward employees for working longer hours, or don’t require them to do so. If employees aren’t familiar with the company’s overtime pay policies, inform them about this. If the expectations around workload and time off aren’t explicit, make them clear. Hold company-wide and within-team discussions to understand and address the company work culture, across the board and in specific contexts. High evaluation scores should not be reserved for employees who work 200+ hours per month. Consider a proportionate scale for rewarding employees who contribute more/put in more effort.