package oplsr

You can search for identifiers within the package.

in-package search v0.2.0

oplsr
- Oplsr
  - PLS
  - Utls

Legend:
Library
Module
Module type
Parameter
Class
Class type

val optimize : bool -> string -> int -> int * float

(nb_comp_opt, r2) = optimize debug train_csv_fn nb_folds Optimize a Partial Least Square (PLS) model. Especially, the optimal number of components is found, as well as the corresponding R^2 (a model regression performance metric in 0:1; near zero is dangerous, near one is good). debug is a verbosity flag. train_csv_fn is the name of the CSV file holding your data. This file must have the value to model in the first column (Y), all other columns are feature values (X_i). This file must be in space-separated dense format. Its first line is a CSV header (column numbers are fine). nb_folds is the number of folds of cross validation; five or ten is standard for this one. (nb_folds > 1) is mandatory.

val train : bool -> string -> int -> string

trained_model_fn = train debug train_csv_fn nb_comp_opt train a model using the given (optimal) number of components. The filename where the model is stored is returned.

val predict : bool -> int -> string -> string -> float list

predictions = \ predict debug nb_comp_opt trained_model_fn test_csv_fn predict using a trained model stored in trained_model_fn, given the optimal number of components nb_comp_opt, reading data to predict from test_csv_fn. The raw list of predicted values is returned.

val predict_to_file : bool -> int -> string -> string -> string -> unit

predict_to_file debug nb_comp_opt trained_model_fn test_csv_fn output_fn predict using a trained model stored in trained_model_fn, given the optimal number of components nb_comp_opt, reading data to predict from test_csv_fn and writing predictions to output_fn.