In this example giving 400 samples looks like there is an accuracy of 91.5% while training beyond that point, accuracy is getting decreased. Are 400 samples enough to keep for this example? If that is the case, we don't care about new entries? What about new unseen data which now are historical data with class assigned? What's your opinion?
Aravinda C V
https://machinelearningmastery.com/how-to-stop-training-deep-neural-networks-at-the-right-time-using-early-stopping/
0 votes
0 thanks
Stavros Koureas
Nice article Aravinda C V , although it describes Keras implementation inside the framework for Neutral Networks. On the other hand, implementing the same technique without another library may give more freedom also for DecisionTreeClassifier, RandomForestClassifier or KNNClassifier.
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training
More Stavros Koureas's questions See All
Similar questions and discussions