Now lets make ten cross validation directories in the working directory and copy the `data.csv` and `sisso.json` into them and run separate calculations for each run.
...
...
@@ -339,7 +333,6 @@ A full example of the testing set output file is reproduced below:
<details>
<summary>The test data file cv_0/models/test_dim_2_model_0.dat</summary>
@@ -388,7 +380,9 @@ To visualize these results we will also use `plot_validation_rmse` at the end, a
Here is an example of the `plot_validation_rmse` output:
<details>
<summary> Cross-Validation results </summary>
</details>
These initial results, particularly the high standard error of the mean for the 1D and 3D models, indicate that more cross-validation samples are needed (Note: you will have different values as the random samples will be different), so lets increase the total number of samples to 100, and redo the analysis
...
...
@@ -417,7 +411,8 @@ As can be seen from the standard error measurements the results are now reasonab
Because the validation error for the three and four dimensional models are within each others error bars and the standard error increases when going to the fourth dimension, we conclude that the three-dimensional model has the ideal complexity.
...
...
@@ -433,7 +428,8 @@ To see the distributions for this system we run
<details>
<summary> Distribution of Errors </summary>
</details>
One thing that stands out in the plot is the large error seen in a single point for both the one and two dimensional models.
...
...
@@ -596,7 +592,7 @@ The result of which is shown below:
<details>
<summary> Final 3D model </summary>
</details>
Additionally you can generate a output the model as a Matlab function or a LaTeX string using the following commands.
...
...
@@ -610,9 +606,9 @@ Additionally you can generate a output the model as a Matlab function or a LaTeX
A copy of the generated matlab function is below.
<details>
<summary> Final 3D model </summary>
<summary> Matlab function of the Final 3D model </summary>
inally, besides regression problems, `SISSO++` can be used to solve classification problems.
As an example of this we will adapt the previous example by replacing the property with the identifier of if the material favors the rock-salt or zinc-blende structure, and change the calculation type to be `classification`.
It is important to note that while this problem only has two classes, multi-class classification is also possible.
...
...
@@ -10,7 +10,6 @@ Here is the updated data file, with the property `E_RS - E_ZB (eV)` replaced wit
<details>
<summary>Here is the full data_class.csv file for the calculation</summary>
Additionally the model files change to better represent the classifier.
The largest changes are in the header, where the coefficients now represent the linear decision boundaries calculated using support-vector machines (SVM).
...
...
@@ -200,7 +199,6 @@ The estimated property vector in this case refers to the predicted class from SV
<details>
<summary>models/train_dim_2_model_0.dat</summary>
```csv
# [(EA_B * Z_A), (r_sigma + r_p_B)]
# Property Label: $$Class$$; Unit of the Property: Unitless
@@ -332,11 +329,14 @@ These changes are a result of different SVM libraries leading to slightly differ
<summary>`SISSO++` Classification </summary>
</details>
<details>
<summary> sklearn SVM </summary>
</details>
However as we decrease the value of `c` an increasing number of points becomes miss classified, suggesting the model is potentially over-fitting the data .
