Model-Improvement

Improving a Classififcation Model

First Prediction : Baseline Prediction

First Model : Baseline Model

Improving a Machine Learning Model depends on two Perspective :

A.From a Data Perspective

• Could we Collect more Data ? (Generally More Data, More Training, More Patterns, Better Accuracy and Better Model)

• Could we Improve our Data ? (Data Cleaning, Adding New Feature or Remove Unnecessary Feature : Dimensionality Reduction)

B.From a Model Perspective

• Is there any Better Model we can use for Training our Data ? (Try Estimators Like Random Forest Classifier)

• Could we Improve the Current Model ? (Improving Parameters and Adjusting Hyperparameters)

Parameters vs Hyperparameters

Parameters : Model Find this Patterns in Data. (By Default Parameters Value)

Hyperparameters : Setting Parameter Values we can Adjust to Improve its Ability to Find Patterns.

Ways to Adjust Hyperparameters :

A.Adjust Manually

B.Random Search Cross Validation

C.Grid Search Cross Validation