fitting

Fitting is a class to generate and store the fitted results. This is the class for generating fitting_results.json

Author: Zeyu Deng Email: dengzeyu@gmail.com

class kmcpy.fitting.Fitting

Main class for model fitting

add_data(time_stamp, time, keci, empty_cluster, weight, alpha, rmse, loocv)

Add data to the Fitting object

Parameters:

• time_stamp (float) – Time stamp string of the fitting

• time (string) – Human redable date time of the fitting

• weight ([float]) – Weights of each NEB data point

• alpha (float) – Alpha value for Lasso regression

• keci ([float]) – Kinetic effective cluster interactions

• empty_cluster (float) – Empty cluster

• rmse (float) – Root mean square error

• loocv (float) – Leave-one-out cross validation error

Return type:

None

fit(alpha, max_iter=1000000, ekra_fname='e_kra.txt', keci_fname='keci.txt', weight_fname='weight.txt', corr_fname='correlation_matrix.txt', fit_results_fname='fitting_results.json')

Main fitting function

Parameters:

• alpha (float) – Alpha value for Lasso regression

• max_iter (int, optional) – Maximum number of iterations. Defaults to 1000000.

• ekra_fname (str, optional) – File name for E_KRA storage. Defaults to ‘e_kra.txt’.

• keci_fname (str, optional) – File name for KECI storage. Defaults to ‘keci.txt’.

• weight_fname (str, optional) – File name for weight storage. Defaults to ‘weight.txt’.

• corr_fname (str, optional) – File name for correlation matrix storage. Defaults to ‘correlation_matrix.txt’.

• fit_results_fname (str, optional) – File name for fitting results storage. Defaults to ‘fitting_results.json’.

Returns:

Predicted E_KRA; DFT Computed E_KRA

Return type:

y_pred (numpy.ndarray(float)),y_true (numpy.ndarray(float))