Training Set
- A training set is the data used to teach a machine learning model how to recognize patterns and make predictions.
- Models learn from the training set to classify or predict outcomes for new, unseen data.
- The size, diversity, and representativeness of the training set affect model accuracy.
Definition
Section titled “Definition”A training set is a collection of data used to teach a machine learning model to recognize patterns and make predictions. It is the dataset the model uses to learn how to classify or predict outcomes for new data.
Explanation
Section titled “Explanation”The machine learning model inspects examples in the training set and learns relationships between input characteristics (for example, size, shape, color, age, gender, income level, or tenure) and the corresponding labels or outcomes. Once trained, the model applies what it learned from the training set to classify or predict outcomes for new instances it has not seen before.
Examples
Section titled “Examples”Example 1: Fruit classification
Section titled “Example 1: Fruit classification”A training set of 1000 different fruits (including apples, oranges, bananas, and pears) is used to teach a model to classify fruit by characteristics such as size, shape, and color. The model might learn that apples are typically round, red or green, and medium in size, while bananas are long, yellow, and medium in size. Presented with a fruit that is round, red, and medium in size, the model will likely predict that it is an apple.
Example 2: Customer churn prediction
Section titled “Example 2: Customer churn prediction”A training set of customer data (including age, gender, income level, and how long they have been a customer) is used to predict whether a customer will churn (cancel their subscription or service). From this training set, the model could learn that customers who are younger and have a lower income level are more likely to churn, while customers who are older and have a higher income level are less likely to churn. The model can then predict churn for new customers.
Notes or pitfalls
Section titled “Notes or pitfalls”- The size and quality of the training set can greatly impact the model’s performance.
- A larger, more diverse training set will typically result in a more accurate model, while a smaller or less diverse training set may lead to a less accurate model.
- It is important to carefully curate the training set to ensure it is representative of the data the model will encounter in the real world.
Related terms
Section titled “Related terms”- Machine learning model
- Classification
- Prediction
- Churn (cancel their subscription or service)