From optimizing operations to reducing risks, businesses across the world are solving difficult problems and discovering new opportunities using artificial intelligence and machine learning. AI/ML has enabled businesses to perform tasks on a scale and scope that had been deemed impossible to achieve. It has increased both speed and accuracy for calculations and predictions. Doctors are using ML for accurate diagnosis and businesses using it for predicting trends and speeding up decision-making. AI/ML is, especially, an integral part of CredoHire’s business model as well. But before describing how AI/ML shaped CredoHire, let’s brush up on our basics of what AI/ML really is.
Artificial intelligence, as the term suggests, is not possessed by humans; rather it is the ability of computers to learn and perform tasks that we usually believe only human intelligence is capable of. Machine learning, a subset of AI, involves making a computer capable of learning without being actually programmed. By creating ML models based on different kinds of algorithms, we essentially feed the computer diverse and good-quality data to train it to show certain outputs on its own. After training it, we test the machine by giving it data it hasn’t seen before and see if the output it gives matches the output we expected from it.
Now that we have the basics covered, let’s talk about how we at CredoHire used AI/ML to solve our business problems. CredoHire is an AI-powered recruitment platform for startups. We provide startups with candidates that are the best fit for their company and the role being offered. But currently, there is no proper way to find out if the candidate will be a good fit for the startups. Various competitive websites in the market provide assessments and give recommendations but they don’t tell if the candidate is a good fit or not. So our problem was to find a solution or a metric to determine and tell the startups whether the candidate looking for recruitment was really a good fit or not. Here comes the role of AI/ML.:
To solve this problem we, at CredoHire, used Machine learning algorithms to predict the fitment of candidates in the startups based on various parameters. We used historical data of various candidates collected from various HRs. The data contained details about their skills, aptitude, and fitment in the startups. We then used a multiple regression model to predict the outcome; that predicted outcome is now used in our candidate reports.
What is a multiple regression model, you ask? A multiple linear regression model explains the relationships between one continuous dependent variable (y) and two or more independent variables(x1,x2,x3………xn).
To implement the regression model, we collected the data and checked the relations between dependent and independent variables using scatter plots and correlations. Now, while doing multiple regression, two possible problems that may arise are Overfitting and Multicollinearity. If there are correlations or redundancy in the data, it might affect the accuracy of the model. So, after handling overfitting and multicollinearity, we implemented our model on the training data and then tested it. And in the end, we successfully got what we were trying to predict with good accuracy.
