I have 3 independent variable which is height(100,300,500),angle(0,30,......,180) and pressure point(1,2.......,112) and each pressure point has 30,000 values of wind pressure coefficient,I want to predict the unknown data by using machine learning algorithm.so which regression model is best suit for this.
AND if consider only 1 dependent variable i.e mean of 30,000 value then which algorithm is best for this case?
I would suggest you to use PLS. But here you have the problem of balancing the 4 blocks (3 independent blocks/vectors and 1 dependent variable block), as this is a data-fusion approach. If the low-level data fusion does not work, this is, simply joining one after another (of course autoscaling and preprocessing the blocks independently before), I would suggest to try a mid-level data fusion. This is, apply a PCA on the 30,000 dependent variables and extract the PCs that explain most of the variance, then join the scores of these PCs together with the 3 independent variables (autoscaling), this way the blocks should be more balanced and the PLS would perform better.
If the simple approaches do not work, then look for something more complex, but try simple first.
I'd like to hear an update about the results, good luck! :)
Shushank Sengar, sometimes, statistical terminology varies across disciplines, but after reading your question, I wonder if you really have 30,000 observations (or values, as you said) of one DV (wind pressure coefficient).
hmmm... a regression model, by definition, have only one DV... does the 30k values are all in a column?
Hi Jonatan Dos Santos Franco. A univariate regression model has only one DV, but a multivariate model has 2 or more DVs. So it is theoretically possible that Shushank Sengar has 30,000 DVs. I still think it is more likely that there are 30,000 observations of one DV. But we won't know for sure until Shushank Sengar clarifies. ;-)
Article Multifarious terminology: Multivariable or multivariate? Uni...
Bruce Weaver it's possible using a manova, right? so he needs to consider that the DV's are correlates in some way.
Jonatan Dos Santos Franco, if there really are multiple DVs, then yes, some kind of multivariate model could be used. Whether it would be called MANOVA vs MANCOVA vs multivariate regression vs something else would depend on the nature of the explanatory variables (and the conventions in one's area of study). But we still need clarification from Shushank Sengar about whether there really are multiple DVs vs multiple observations of one DV.
I am giving you dataset for which the code needs to be design.Please send the best algoritm which gives the best accuracy for the given code.
Jonatan Dos Santos Franco Bruce Weaver Jokin Ezenarro
Shushank Sengar that is not the complete dataset, I would need the 30,000 variables and also the value to be predicted (the Y-block).
Hello Shushank Sengar. I suppose each value in the "mean value" column is the mean of 30,000 raw wind pressure scores, is it? If so, I would say you have one DV with 30,000 repeated measures for each combination of the 3 explanatory variables.
Shushank Sengar i agree with Bruce Weaver , you have a *one DV* with 30,000 observations, BUT, don't is 30,000 repeated measures, in fact, you have three repeated measures: 100, 300 and 500 (heights), for each value of heights you have 7 values of degree, right? 0, 30, 60, 90, 120, 150 and 180. So, you need to edit you database adding a 'time' column.
I think that your model need to be like that:
Y(mean_value) = B0(t) + b1*degree(t) + b2*height(t) + b3*pressure(t) + b4*time + e
You want to predict the mean value from the IV, but you need to do this with a repetead measure model (GEE or GMM).
Shushank Sengar, was the data file you shared the full dataset? Or was it a subset of a larger data file that has 30,000 rows? Basically, I think everyone is still uncertain about where the 30,000 you talk about comes in. As you saw in my earlier post, I speculate that the "mean value" in the final column is the mean of 30.000 raw wind pressure measurements. But Jonatan disagrees. Only you can clarify this. Thanks.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python