The important aspect of H-methods is that each of them provides with graphic procedures that show the inherent variation in data. There are typically four types of graphics that can be carried out.

1. Computed values and residuals
The graphic procedures here are among others:
a) Observed versus computed Y-values. The columns of Y are drawn against the corresponding columns of Y_est=XB_A. The graphs are supplied by different measures of how good a fit has been obtained.
b) Y-values against the score vectors. These graphs show the quality of the fit at each step of the computations.
c) Y--values against the residuals. The residuals are given as E=Y-XB_A. If the plots of the columns of Y against the corresponding column of E show systematic variations, it indicates that the modeling task has not been successful.  
d) The Y-residuals. The columns of the residual matrix E are to exhibit random behavior. Therefore, plots, where the y-axis is a column of E and x-axis is e.g., the sample number, a score vector or a variable, should show random scatter of points.

2. Results of optimization at each step
The predicition aspect of the model is optimized at each step. The results can be graphically illustrated as:
    a) Plots of the resulting weight vector
    b) Size of the covariance
    c) Eigen values found

3. Plots of vectors
At each step there are computed
    a) weight vectors, w_a, which show the importance of the variables,
    b) score vectors, t_a=X_a-1w_a, showing the variation in the samples
    c) loading vectors, p_a=X_a-1^Tt_a, showing the correlation/covariance with the score vector
    d) scaling constants, d_a=1/(t_a^Tt_a),  showing the 'precision' at each step
    e) loading weight vectors, v_a, computed as, t_a=Xv_a, showing how the original variables generate the score vectors

4. Model control
This is carried out in several different ways
a) Divide the samples into ten parts of equal size. For each of the ten parts use the results from applying the procedure to the other nine to evaluate its performance at the part.
b) Bootstrapping