I have real time datasets of 3 years dealing with energy management of households. Now these datasets are not enough to conclude that my model will perform well on every situation. For that, i need to create more datasets based on the behavior of previous years datasets. Can anyone provide any insights, how can i create more datasets artifically while creating different scenarios in MATLAB?
Hope this will help :
https://www.mathworks.com/matlabcentral/fileexchange/41459-6-functions-for-generating-artificial-datasets
It is a bit unclear what you really want to do.
If your artificial datasets have to be BASED ON EXISTING data.
Don't you just need a resample procedure by basically picking random cases in your Ntot data ?
https://www.mathworks.com/help/stats/datasample.html
If you have a model based on this 3 years and you want to infere its predictive power, then you need data OUT of the training data.
For this, you shall train your model on n < Ntot points and you can have a look at the bootstrap procedure.
https://machinelearningmastery.com/a-gentle-introduction-to-the-bootstrap-method/
Muhammad Ali One way could be synthetic data generation. You can Google about it. You can try to find the statistical model (e.g.: normal, gaussian etc.) that match with your previous collected valid data. And then, you can use Matlab to generate synthetic data with that statistical model.
(Caution: It will be hard to prove the validity of this data generation process / to be precise that this process actually reflects similarity with your dataset)
Type of data?
1)Text; open file.
2) Measured; computer interfacing.
As seeing on work space save and load as required.
A. K. M. Sharoar Jahan Choyon .The attached link is for generating syntheic data for imagaes. My data is in numberic text e.g columns with values like 0.245789
Eanass Al-Shabkhoon My data is in numeric digits (.csv) e.g columns with values like 0.245789
Atanu Shome .I agree with you. Can you recommend any useful tutuorial in MATLAB dealing with generation of syntheic data?. My data is in numeric digits (.csv) e.g columns with values like 0.245789 .
Damien Mercier Thank you for very useful answer. Actually, I have real time datasets of 3 years dealing with energy management of households. Now these datasets are not enough to conclude that my model will perform well on every situation. For that, i need to create more datasets based on the behavior of previous years datasets. Can anyone provide any insights, how can i create more datasets artifically. I have uploaded the data sample here: https://filebin.net/vfantyz184i3l6t2
It must be return to MATLAB help at any way the are three step:
1_ fopen the instruction of opening file.
2_ fread the instruction of reading.
3_fclose the instruction closing file.
Try to see the utilities behind these instructions, good luck.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering