06. Training and test sets

Training and test sets

---

Topic: Training and test sets

Course: GMLC

Date: 17 February 2019

Professor: Not specified

---

Resources

---

• https://developers.google.com/machine-learning/crash-course/training-and-test-sets/video-lecture

• https://developers.google.com/machine-learning/crash-course/training-and-test-sets/splitting-data

• https://developers.google.com/machine-learning/crash-course/training-and-test-sets/playground-exercise

Key Points

---

• In case of having a single data set, it is recommended to split the data set into a test  and training  set

• It is important to remove duplicates and randomize the data set before splitting, otherwise we might get a falsely low loss (accidentally train on test data)

• Test data characteristics

  • Large enough to yield meaningful results

  • Is representative of the dataset as a whole (randomize before splitting)

Check your understanding

---

• Know test data characteristics

• Know what is important before splitting the data

Summary of Notes

---

• We can create 2 subsets from one dataset by splitting it

• It is important to randomize & remove duplicates before splitting