Exploratory analysis and predictive modeling for a two-sided service marketplace

Goal Predict hiring decision by users
Tools SQL, Python (scikit-learn)
Data Sources Internal

Have you ever had to hire a plumber? How did you find and hired one? This project was requested by an upcoming freelance and service marketplace which was eager to learn what drives their user decision when it comes to hiring and what are the relevant factors affecting the decision

The model seems to be able to predict positive and negative hires decently well. The confusion matrix shows good accuracy on predicting hires (93%) and even better at spotting negatives (97%).

In order to calculate the relative importance of an attribute in the model, we have calculated the magnitude of its coefficient times the std dev of the corresponding parameter.

It seems that number of impression in search, position, number of previous hires and and hourly rate are more relevant at predicting hired.

Visit GitHub Repo