Model Fitting

What is Model Fitting?

Fitting is a measure of how well a machine learning model generalizes to similar data to that on which it was trained. A model that is well-fitted produces more accurate outcomes, a model that is overfitted matches the data too closely, and a model that is underfitted doesn’t match closely enough.

Each machine learning algorithm has a basic set of parameters that can be changed to improve its accuracy. During the fitting process, you run an algorithm on data for which you know the target variable, known as “labeled” data, and produce a machine learning model. Then, you compare the outcomes to real, observed values of the target variable to determine their accuracy. Next, you use that information to adjust the algorithm’s standard parameters to reduce the level of error, making it more accurate in uncovering patterns and relationships between the rest of its features and the target. You repeat this process until the algorithm finds find the optimal parameters that produce valid, practical, applicable insights for your practical business problem.

Why is Model Fitting important?

Fitting is the essence of machine learning. If your model doesn’t fit your data correctly, the outcomes it produces will not be accurate enough to be useful for practical decision-making (keeping in mind the dangers of overfitting). A properly fitted model has hyperparameters that capture the complex relationships between known variables and the target variable, allowing it to find relevant insights or make accurate predictions.

Fitting is an automatic process that makes sure your machine learning models end up with the individual parameters best suited to solve your specific real-world business problem with a high level of accuracy.

Model Fitting + DataRobot

DataRobot automatically fits dozens of models on your data simultaneously, drastically reducing the amount of time it would ordinarily take. Also, while fitting itself is an automatic process, more complicated data science techniques that increase model accuracy like tuning hyperparameters ordinarily require a significant amount of time and data science experience. DataRobot automates this work so that so any model can be fitted with the least possible effort.