ML Model Development

With the experience and expertise we have at hand, we maintain a structured and repeatable process for every machine learning problem. This helps us to deliver the best possible results in the least amount of time.

We use a multitude of tools and visualizations to assist with the developmental stages of our projects.


This helps us to initially determine which features correlate with other features or with the target variable. This visualization allows us to make initial conclusions as to which features are likely to have the most output on the model’s predictions. If two features are highly correlated, we may want to drop one of these.

Correlation Matrix

This explains to us how the model is making its predictions. From this plot, we can determine which features are most and least significant to the model’s output and can ascertain which direction the prediction is moved by higher or lower values of each feature.

Model Explainer - SHAP

This indicates whether a model’s predictions are improving with the addition of each new feature. Similar plots can be used to examine F1 scores, RMSE or MAE.

Plot of Accuracy with Addition of New Features

This gives an overview of the hyperparameter tuning process. The values of each parameter can be directly compared to their respective effect on the performance of the model. This allows us to pick the best choice of hyperparameters.

Parallel Coordinates Plot