xref: /petsc/src/ml/regressor/tests/ex_sharks.c (revision c12c126234ed623246a63bfa78c9f75a3aa00323)
1 /* Example inspired by the toy example in https://www.r-bloggers.com/2020/06/understanding-lasso-and-ridge-regression-2/
2  * blog post by Dr. Atakan Ekiz.
3  * Here we wish to predict the number of shark attacks (that is, this number is our response variable),
4  * using the following predictor variables:
5  * - percentage of swimmers who watched the movie Jaws
6  * - the number of swimmers in the water
7  * - the average temperature of the day
8  * - the price of your favorite tech stock of the day (totally uncorrelated variable) */
9 
10 static char help[] = "Tests basic creation and destruction of PetscRegressor objects.\n\n";
11 
12 #include <petscregressor.h>
13 
main(int argc,char ** args)14 int main(int argc, char **args)
15 {
16   PetscRegressor regressor;
17   PetscMPIInt    rank;
18   Mat            X;
19   Vec            y, y_predicted, coefficients;
20   PetscScalar    intercept;
21 
22   PetscScalar y_array[20] = {98, 53, 39, 127, 73, 42, 71, 61, 83, 74, 85, 82, 62, 60, 43, 69, 67, 69, 85, 3}; // Number of shark attacks
23 
24   PetscScalar X_array[80] = {37.92934, 513, 92.89899, 137.2139, // % watched Jaws, #swimmers, temperature, stock price
25                              52.77429, 451, 87.86271, 145.7987, //
26                              60.84441, 456, 88.28927, 149.7299, //
27                              26.54302, 546, 89.43875, 147.1180, //
28                              54.29125, 431, 88.01132, 124.3068, //
29                              55.06056, 355, 88.06297, 114.1730, //
30                              44.25260, 557, 87.78536, 112.5773, //
31                              44.53368, 398, 87.49603, 125.1628, //
32                              44.35548, 498, 88.95234, 124.8483, //
33                              41.09962, 406, 89.00630, 115.9223, //
34                              45.22807, 610, 86.38794, 148.1111, //
35                              40.01614, 452, 88.83585, 131.7050, //
36                              42.23746, 429, 87.78222, 106.3717, //
37                              50.64459, 450, 87.97008, 121.1523, //
38                              59.59494, 337, 89.67538, 145.7158, //
39                              48.89715, 383, 91.12611, 123.3896, //
40                              44.88990, 282, 93.29563, 145.4085, //
41                              40.88805, 366, 88.45329, 129.8872, //
42                              41.62828, 471, 93.21182, 131.5871, //
43                              74.15835, 453, 87.68438, 143.4579};
44 
45   PetscInt rows_ix[20] = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19};
46   PetscInt cols_ix[4]  = {0, 1, 2, 3};
47 
48   PetscCall(PetscInitialize(&argc, &args, (char *)0, help));
49   PetscCallMPI(MPI_Comm_rank(PETSC_COMM_WORLD, &rank));
50 
51   PetscCall(VecCreate(PETSC_COMM_WORLD, &y));
52   PetscCall(VecSetSizes(y, PETSC_DECIDE, 20));
53   PetscCall(VecSetFromOptions(y));
54   PetscCall(VecDuplicate(y, &y_predicted));
55   PetscCall(MatCreate(PETSC_COMM_WORLD, &X));
56   PetscCall(MatSetSizes(X, PETSC_DECIDE, PETSC_DECIDE, 20, 4));
57   PetscCall(MatSetFromOptions(X));
58   PetscCall(MatSetUp(X));
59 
60   if (!rank) {
61     PetscCall(VecSetValues(y, 20, rows_ix, y_array, INSERT_VALUES));
62     PetscCall(MatSetValues(X, 20, rows_ix, 4, cols_ix, X_array, ADD_VALUES));
63   }
64   PetscCall(VecAssemblyBegin(y));
65   PetscCall(VecAssemblyEnd(y));
66   PetscCall(MatAssemblyBegin(X, MAT_FINAL_ASSEMBLY));
67   PetscCall(MatAssemblyEnd(X, MAT_FINAL_ASSEMBLY));
68 
69   PetscCall(PetscRegressorCreate(PETSC_COMM_WORLD, &regressor));
70   PetscCall(PetscRegressorSetType(regressor, PETSCREGRESSORLINEAR));
71   PetscRegressorSetFromOptions(regressor);
72   PetscCall(PetscRegressorFit(regressor, X, y));
73   PetscCall(PetscRegressorPredict(regressor, X, y_predicted));
74   PetscCall(PetscRegressorLinearGetIntercept(regressor, &intercept));
75   PetscCall(PetscRegressorLinearGetCoefficients(regressor, &coefficients));
76 
77   PetscCall(PetscPrintf(PETSC_COMM_WORLD, "Intercept is %lf\n", intercept));
78   PetscCall(PetscPrintf(PETSC_COMM_WORLD, "Coefficients are\n"));
79   PetscCall(VecView(coefficients, PETSC_VIEWER_STDOUT_WORLD));
80   PetscCall(PetscPrintf(PETSC_COMM_WORLD, "Predicted values are\n"));
81   PetscCall(VecView(y_predicted, PETSC_VIEWER_STDOUT_WORLD));
82 
83   PetscCall(PetscRegressorDestroy(&regressor));
84 
85   PetscCall(PetscFinalize());
86   return 0;
87 }
88