Exploratory workflows

The four exploratory tools take the same data shape — samples down the rows, variables across the columns, an optional label column and an optional group column — and each produces its own set of figures and tables. What the methods mean is in the PCA vs PLS, correlation and heatmap guides; this page is the operational path.

Getting the data in

Paste a sample × variable table

Workbench → pick the tool → paste the table → tick first column = sample names and/or a group column if present → Confirm

The preview lists what was parsed — samples, variables, groups. For PLS, mark the response column when asked.

PCA

Run and read

Preview data → Confirm → Run PCA → the variance table, the score plot and the biplot appear

The score plot shows samples (coloured by group if you gave one); the biplot adds the variable arrows. The variance table says how much each component explains; the loadings table which variables drive each.

Choose components and scaling

Options → Components → Scale variables on/off

Scaling (the default) puts variables on equal footing; turn it off only when they share units and their variances are meaningful.

PLS

Run and read

Preview data → Confirm (with the response marked) → Run PLS regression → variance per component, the observed-vs-predicted plot, the importance (VIP) and coefficients tables

The observed-vs-predicted plot with its cross-validated R² tells you whether the model predicts; the VIP table which variables matter.

Components, cross-validation, permutation

Options → Components / Cross-validation / Permutation test

Cross-validation guards against too many components; the permutation test asks whether the fit beats chance.

Correlation matrix

Run and read

Preview data → Confirm → Method (Pearson, Spearman, Kendall) → the run button → the coloured matrix and the r / p table

Cells are coloured by the coefficient; the table beneath gives every pair’s r and p.

Heatmap

Run and read

Preview data → Confirm → the run button → the clustered heatmap with its trees

Rows and columns are clustered (Ward linkage on standardised values by default); the trees show the merge order. The Trees and Colour bar tabs of Figure styling control their look.

Add a group strip

Give a group column in the data → the strip appears beside the rows

The strip colours each sample by its group without affecting the clustering.