Constructing a prioritization score from the following ordinal variables:

Last Visit [100,80,60,40,20,0]
Num Emp [100,0]
Year Established [100,80,60,40,20,0]
Last Exp [100,50,0]
Tier [100,66,33,0]

listed in order of importance. For now, importance is simply a multiplier to the score, ranging [1,5].

These are nullable (num_emp has many nulls).

I understand I need to transform them. A MinMaxScaler has been used, but an omitted variable(s) will artificially bias the prioritization score, given the differing number of ordinals.

No pairs of variables can be assumed to be bivariate normal (all but num_emp are time-series and right-skewed).
tier cannot assume an underlying latent continuous variable.
As such, a polychoricpca cannot be performed.

Creating dummy vars will not preserve the ordinal characteristics of the variables.

I’m not sure if PCA is appropriate, as the coefficients are given. I am most concerned with properly transforming the categorical variables.