We were doing a project to integrate two different technologies, Civil Engineering and Machine learning in my project. We have completed till the dataset preparation and data cleaning steps.
Now, I have a challenging task to build a model for this data. But our data has 9 independent variables and 3 target attributes. Each target attribute correlates with all the 9 inputs. And also 3 target attributes correlate each other.
I am not getting which algorithm/model I should apply for this data. It will be so helpful if anybody suggest me or give me an idea. Here I am attaching a small sample of our data .
Hi,
To begin with, try to understand the correlation between each independent variable and target variable. This can be done using graphs or correlation matrix and based on your experience in the subject. Say, you observed a linear relationship, then build a linear model. If you observe exponential or higher degree relationship, then build one accordingly for each target variable.
In case, there are models existing in the literature which are similar to your study, then take them as reference and try different possibilities using tools like cftool in MATLAB or SPPS.
If multilinear regression isn't giving a reasonable prediction, then try using ANN (linear and non-linear transfer functions).
I hope this helps you.
Regards
Kiran.
Thank you sir.
After taking suggestions from Kiran sir, I understood the behaviour of inputs and outputs is linear. So I tried using multi output linear regression model.
I will work on your suggestion also sir. And compare the accuracy of both.
Thank you sir
- Badges
- Science topic