Published in

Simple AI

# How Good is Your Model? — Intro To Machine Learning #4

Hi, this is the Third Article on our journey about Machine Learning Algorithms, You can find here Part 1, Part 2, Part 3 (depending on your background that might not be required for reading this article).

In this post I’m going to show you two simple ways you can use to evaluate your machine learning model.

# Agenda:

• Why model evaluation?
• Problems with training and testing on the same data
• Train / test Split Method
• Cross validation method
• Comparing cross-validation to train/test split

# Why Model Evaluation

Whenever you have problem and you want to solve it using machine learning, one thing you will have to ask yourself after choosing your machine learning model is: how good or bad is your model? The reason for that is to know how it will perform when used to make predictions on unseen data.

# Problems with Training and Testing on the same Data

Goal is to estimate likely performance of a model on out-of-sample data, But, maximizing training accuracy rewards overly complex models that won’t necessarily generalize, Unnecessarily complex models overfit (also called over-learning) the training data, overfiting happens when the model learn specific details in noise data.

Lets take the image above as an example, the lines represent decision boundaries, the lines separate the positive examples (red) from negative examples (blue).

The green line is likely to perform poorly on out-of-sample data (unseen data) because it learns that noise data points while the black while can be considered the best as it doesn't follow noise data points which is good to generalize the training data. Overfiting is a big deal in machine learning and it can be a bit difficult to understand at first but you will get it.

# Evaluation Methods

## Evaluation procedure (1): Train/test split

1. Split the dataset into two pieces: a training set and a testing set.
2. Train the model on the training set.
3. Test the model on the testing set, and evaluate how well we did.

## Evaluation procedure (2): K-fold cross-validation

1. Split the dataset into K equal partitions (or “folds”).
2. Use fold 1 as the testing set and the union of the other folds as the training set.
3. Calculate testing accuracy.
4. Repeat steps 2 and 3 K times, using a different fold as the testing set each time.
5. Use the average testing accuracy as the estimate of out-of-sample accuracy.

# Comparing Cross-validation To Train/Test Split

• More accurate estimate of out-of-sample accuracy
• More “efficient” use of data (every observation is used for both training and testing)

• Runs K times faster than K-fold cross-validation
• Simpler to examine the detailed results of the testing process

## Which One Is the Best?

I guess you know the answer for that by now, The Best Method? It depends on your dataset and the computational resources available to you, on large datasets cross validation is computationally expensive to run (trust me, it can be boring).

# Resources:

## Next:

In the next article we are going to talk about Two Types of machine learning problems: Classification and Regression.