How W4M can be used to perform statistics including exploratory data analysis, hypothesis testing, machine learning and feature selection ?
At the end of the course, you will be able :
- to view the data (PCA, heatmap) ;
- perform statistical tests and apply corrections for multiple testing ;
- build predictive models (PLS, Random Forest, SVM) ;
- select the variables which are signifcant for the predictive model.
Basic knowledge of biostatistics and multivariate data analysis.
Here, we describe how to analyze a ‘sample by variable’ table of intensities, such as the one we generated during the previous ‘processing’ step. The objective is to explore the data (e.g. detect trends, clusters, or outliers), perform univariate hypothesis tests, build predictive models for the factor of interest (regression or classification), and select the significant variables (i.e. the molecular signature) for robust and high performance.
- Exploratory data analysis
- Hypothesis testing
- Multivariate predictive modeling
- Feature selection
W4M allows you to build comprehensive and reproducible workflows for data analysis. Diagnostics and correction methods are included to correct for multiple testing and avoid overfitting. The available modules can be applied to targeted or untargeted omics data.