Timeline for How to test dev set on Time Series data via forecasting
Current License: CC BY-SA 4.0
3 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
Jun 24, 2019 at 18:42 | comment | added | Skiddles | Thanks for following up. I was curious if my answer was helpful. One thing I will say based on your response, 10k samples is not particularly small, and more data doesn’t necessarily mean more predictive power. It may lead to a more generalized model which may or may not be beneficial. Good luck to you. | |
Jun 24, 2019 at 18:36 | comment | added | thiago_lira | Thanks for the valuable insight! Since I have a small dataset (10k training samples) I was using a small number of epochs, but increasing the number has indeed helped. I'm also now fixing a seed now on all my models. | |
Jun 18, 2019 at 17:48 | history | answered | Skiddles | CC BY-SA 4.0 |