From the course: DevOps for Data Scientists
Join today to access over 22,600 courses taught by industry experts or purchase this course individually.
Testing and validating models
From the course: DevOps for Data Scientists
Testing and validating models
- [Instructor] Predictive models are useful only when their predictions are sufficiently accurate as to improve decision-making. An important part of data science practice is understanding how well a model works. So for example, flipping a coin will probably make accurate predictions of binary questions 50% of the time. We can expect our models to perform significantly better than that. A common practice in machine learning and data science is to keep a subset of training data as a test set. This test set data is not used to build the models, but once the model is built, we use the test data to evaluate how well the model works. We use test set data during model development. After we deploy a model to production, we still need to continue to validate or test it. Patterns in the data may change over time, and the model may begin to drift or become less accurate because the data used to train the original model is no longer representative of the data that we see in production. So to…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.