Hi everyone! I am modeling my dataset to predict the Temperature and Light(New Value) in different locations (Translate02). However, I am a bit confused about how to configure the nodes. May I ask for help?
setc.knwf (45.5 KB)
If your data is not confidential, can you post that as well?
What specific questions about configuration of the nodes do you have?
1 Like
Hi, sure. The predicted value and its original value are quite different. However, the r2 in scorer shows it is accurate enough which is 0.8. I am very confused that which step I did wrongly result in this outcome. by the way, I thought my dataset is already inside the workflow?
Hello @yukiii1223,
Nop. Workflow is in state reset and data is read from your local drive. To have data with your workflows you can either not reset workflow when exporting it (reset is default) either place file in data folder within workspace location and use workflow relative path.
Br,
Ivan
hihi this is my dataset. Sorry for the delay
Hello @yukiii1223,
I love to play with all kind of data, so i hope you don’t mind if I step’in.
I have toyed with your dataset and came up with something “not bad” :
For Light prediction :
For Temp prediction :
Max, Mean, Skewness and Standard dev. are not bad. Min is not very accurate so there is room for improvement.
I added a column “discrepancies” to check the difference between the actual and the predicted.
Find attached the workflow I made. i hope it’ll help you.
setc.knwf (69.1 KB)
I used Random Forest (Regression) because I like to use for classification problem. I think that your predictions could be improved by increasing the dimension reduction (i.e categories or groups of data), maybe removing the outliers.
Notice that i have first cut my data in 2/3 for training, which 80% is for training and 20% for testing as usual, and 1/3 for validation, to be sure that my model didn’t overfit or underfit.
Br,
Samir
2 Likes
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.