Dear all,
I have a dataset from several lakes describing their zooplanktonic communities (species abundances) and I would like to use multivariate methods to analyze the data (I am provisionally oriented to a RDA / CCA).
My question is: how to manage repeated samples? I sampled each lake several times (more specifically from 1 to > 30 times). I am not very familiar with multivariate analyses, but I feel that this unbalanced sampling could cause some problems (something similar to pseudoreplications?!). Can anybody help me? Is it sufficient to include the variable "lake" in the explanatory matrix of the environmental factors?
Thank you
Rocco
Dear Rocco Tiberti ,
You may considered unique commant in matlab to select unique sample after that you should considered outliers analysis to removes the outliers. After removing effective outliers and getting unique sample, you may apply multivariate regression analysis.
If you face any problem then, please mail me your data samples ([email protected])
Please follow these papers:
https://www.researchgate.net/publication/303011477_Quality_Assessment_of_Web_Services_Using_Multivariate_Adaptive_Regression_Splines
https://www.researchgate.net/publication/299328355_Hybrid_functional_link_artificial_neural_network_approach_for_predicting_maintainability_of_object-oriented_software
https://www.researchgate.net/publication/283723707_Predicting_Object-Oriented_Software_Maintainability_using_Hybrid_Neural_Network_with_Parallel_Computing_Concept
https://www.researchgate.net/publication/275071932_Statistical_and_Machine_Learning_Methods_for_Software_Fault_Prediction_Using_CK_Metric_Suite_A_Comparative_Analysis
https://www.researchgate.net/publication/283723964_Hybrid_neural_network_approach_for_predicting_maintainability_of_object-oriented_software
https://www.researchgate.net/publication/283723881_Fault_Prediction_for_Apache_Open_Source_Framework_using_Chidamber_and_Kemerer_Metrics_Suite
https://www.researchgate.net/publication/292677172_Maintainability_prediction_of_web_service_using_support_vector_machine_with_various_kernel_methods
Conference Paper Quality Assessment of Web Services Using Multivariate Adapti...
Article Hybrid functional link artificial neural network approach fo...
Conference Paper Predicting Object-Oriented Software Maintainability using Hy...
Article Statistical and Machine Learning Methods for Software Fault ...
Article Hybrid neural network approach for predicting maintainabilit...
Conference Paper Fault Prediction for Apache Open Source Framework using Chid...
Article Maintainability prediction of web service using support vect...
It depends what you want to do...each lake measurement is different than the other or more of the same? Are the measures you measure in all of the lakes behaves similarly?. In order to understand what to do with the measurements you need to give more information. If you need you can email me [email protected].
Hi,
if the lakes are the same, then you should restructure your dataset. Imagine you have a variable Lake1 which you meausred 10 times. In your dataset you will have caseID 1 with variables Lake1, Lake2, Lake3...Lake10. You go to SPSS, Data, Restructure, and make from 10 variables a single variable "Lake". Your dataset will be restructured in a way, that 1. you will have a new (additional) index varialbe CasedID = 1, IndexID1 = 1 (first measurement), IndexID1 = 2 (for the second measurement) and so on. Then you just include this index variable in your analysis and t will show you whether the time of measurement had influence on your variable "lake".
If you have more then 20 measurement times for your variables, then you could apply panel data anylsis.
Good news: each measurement increases your sample size by 100% ;)
Faithfully,
Eugene
- Badges
- Science topic