The Challenge

Workplace diversity is now something most companies strive to achieve. Companies with diverse workforces make better decisions faster, which gives them a serious advantage over their competitors. As a result, companies with diversity in the workplace achieve better business results and reap more profit.

The aim of this Oil & Gas major was to make the organisation more inclusive and provide equal opportunities to people across diversity groups which would result in numerous benefits across the business.

Chaucer have a longstanding relationship with this Oil & Gas major and have several active projects. This gives us a detailed knowledge of the organisation, culture and technologies allowing us to deep insights and ability and a unique ability to deliver their overarching business objective.

The Solution

In order to achieve this goal, Chaucer suggested the following data opportunity:

Diversity analysis: talent acquisition and career progression predictive modelling.Predictive modelling to leverage Machine Learning methodologies to predict the success of an individual’s application and highlight the reasons for that decision, potentially exposing biases in the application process.

Chaucer hypothesised that the results of the application process, that being internal (as career progression) or external (hiring process), are driven by biases.

The goal of this task was to understand these biases and specifically analyse the importance of 3 diversity attributes:

1. Gender
2. Age
3. Nationality

A Machine Learning model was developed that could predict the success of an individual’s employment application to understand human biases within the decision-making process. The modelling was conducted using a 3-step approach:

1. Build candidate diversity prediction models.
2. Use Natural Language Processing to analyse how job descriptions, both internal and external, affect candidate diversity.
3. Dissect different diversity groups to further understand the driving factors of acquisition/progression decisions.
4. The modelling used rich client datasets to provide a robust assessment of the biases behind the application decision process.

The Results

1. Enhanced Analysis – greater understanding of employee behaviour and use of models by HR to predict positive recruitment strategy
2. Reputational advantage – ensuring organisation can meet its 25% target representation by Q4 2020
3. Increased efficiency - with a large part of analysis already done, and the model continuing to run with updated outputs, efficiency is improved for those involved in analysing the data.