dataset
a group of data kept together and used for study, analysis, or computer programs
A collection of related data, usually arranged or stored so it can be searched, analyzed, or used by software.
- Base: dataset
- Plural: datasets
benchmark dataset
1standard data for comparison
a standard dataset used to compare results fairly
- Their algorithm achieved strong results on a well-known benchmark dataset.
- Benchmark datasets make it easier to compare new models with older ones.
- They tested the model on a benchmark dataset.
- A benchmark dataset helps compare two systems.
labeled dataset
1data with assigned labels
data where each example has a name, category, or correct answer attached
- The team created a labeled dataset of customer emails marked as urgent or non-urgent.
- A labeled dataset is essential for many supervised learning tasks.
- Each picture in the labeled dataset has a category.
- We need a labeled dataset for this task.
open dataset
1freely available data
a dataset that anyone can usually access and use
- The city published an open dataset of bus arrival times.
- Researchers used open datasets to study air quality around the world.
- This open dataset is free to download.
- The website has many open datasets.
real-world dataset
1data from actual situations
a dataset made from real situations, not made-up examples
- The system worked in the lab, but it failed on a messy real-world dataset.
- We evaluated the tool using a real-world dataset from a hospital network.
- They tested the app on a real-world dataset.
- This real-world dataset has many missing values.
synthetic dataset
1artificially generated data
a dataset made by a computer or simulation, not collected from real life
- The company generated a synthetic dataset to test the software without exposing customer information.
- Synthetic datasets can help when real data is rare, expensive, or sensitive.
- They made a synthetic dataset for testing.
- The synthetic dataset did not include real customers.
test dataset
1data for testing models
data used to check how well a model works
- Accuracy was measured on a test dataset that the model had never seen.
- The test dataset included examples from several different countries.
- We used a test dataset to check the model.
- The model scored 92 percent on the test dataset.
training dataset
1data for training models
data used to help a computer model learn
- The model performed well on the training dataset but poorly on new data.
- We need a larger training dataset before the system can recognize rare objects.
- The training dataset helps the model learn.
- They used photos as a training dataset.
validation dataset
1data for model tuning
data used while improving a model before the final test
- We adjusted the learning rate based on results from the validation dataset.
- The validation dataset helped us choose the best version of the model.
- The validation dataset helps us improve the model.
- They checked each model on the validation dataset.