-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_randomforest.m
216 lines (177 loc) · 8.43 KB
/
test_randomforest.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
% This file is part of scikit-from-matlab.
%
% scikit-from-matlab is free software: you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation, either version 3 of the License, or
% (at your option) any later version.
%
% scikit-from-matlab is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details.
%
% You should have received a copy of the GNU General Public License
% along with scikit-from-matlab. If not, see <https://www.gnu.org/licenses/>.
% Author: Abhishek Jaiantilal ([email protected])
% scikit-from-matlab 0.0.1
% This uses RandomForest as an example to show how to call the RandomForest algorithm
% as-is (default parameters), custom values for parameters,
% use a CV strategy and pass custom values for CV parameters
%This is a small test of the scikit_from_matlab package using RandomForest it shows how
%for windows i put in the dll files in a subfolder. seems like that is the
%only way i could run RandomForest. maybe expanding the matlab python path to
%include that will help
clear classes %clears the py classes. seems like 'clear all' doesn't clear these python classes
path_to_package = '.';
scikit_setup(path_to_package); %%%%%%%%%%%%%call this script to set things up <- need to set this up
addpath('data');
num_threads = 8; %change this to how many threads are available
%it seems that both GridSearchCV and RandomForest uses only 1 thread if this variable is not specified
%so make sure to set n_jobs to num_threads (just search for n_jobs in this
%code), i have it complain on an old version of matlab and i had to comment it out
%%%% Classification example
%load twonorm, fudge the Ytrn to add a new class. then send and see if it
%works
load twonorm
Xtrn = inputs';
Ytrn = outputs;
Ytrn(end-10:end) = 3;
Xtst = Xtrn;
Ytst = Ytrn;
%Let's try RandomForest https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html
% Classification with default parameters
algo_name = 'RandomForestClassifier';
fprintf('Tutorial 1: running %s with default parameters\n', algo_name);
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name);
toc
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
accuracy = length(find((ypred ~= Ytst))) / length(ypred) %accuracy
%now we send in some algo_params, lets try with n_estimators=50
fprintf('Tutorial 2: running %s with additional parameter of num estimators = 50\n', algo_name);
%create a dict(ionary) object and pass in the argument. Make sure if its
%int or float or whatever type required in scikit algorithm
algo_params = py.dict;
update(algo_params, py.dict(pyargs('n_estimators',int64(50))))
update(algo_params, py.dict(pyargs('n_jobs',int64(num_threads))))
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name, algo_params);
toc
fprintf('model now has this many n_esimators %d\n', int64(model.n_estimators))
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
accuracy = length(find((ypred ~= Ytst))) / length(ypred) %accuracy
%now we use some form of CV, lets try GridSearchCV, with search over
%lets try max_depth, learning_rate and gamma for RandomForest
fprintf('Tutorial 3: running %s with additional parameter of GridSearchCV and trying (max_features, n_estimators) search\n', algo_name);
algo_params = py.dict;
CV_strategy = 'GridSearchCV';
CV_params_for_algo = py.dict;
update(CV_params_for_algo, py.dict(pyargs('n_estimators', int32(100:200:500) )))
update(CV_params_for_algo, py.dict(pyargs('max_features', int32(1:2:10) )))
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name, algo_params, CV_strategy, CV_params_for_algo);
toc
fprintf('model parameter searched over\n')
model.param_grid
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
accuracy = length(find((ypred ~= Ytst))) / length(ypred) %accuracy
%now we use some form of CV, lets try GridSearchCV, with search over
%lets try max_depth, learning_rate and gamma for RandomForest
%and pass parameters to GridSearchCV
fprintf('Tutorial 4: running %s with additional parameter of GridSearchCV and trying (max_features, n_estimators) search\n', algo_name);
fprintf('Furthermore we now select a CV time of 4 and do refit=true\n')
algo_params = py.dict;
CV_strategy = 'GridSearchCV';
CV_params_for_algo = py.dict;
update(CV_params_for_algo, py.dict(pyargs('n_estimators', int32(100:200:500) )))
update(CV_params_for_algo, py.dict(pyargs('max_features', int32(1:2:10) )))
%https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html
CV_params = py.dict;
update(CV_params, py.dict(pyargs('cv', int64(4) )))
update(CV_params, py.dict(pyargs('refit', int64(true) )))
update(CV_params, py.dict(pyargs('n_jobs', int64(-1) ))) %-1 means all processors
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name, algo_params, CV_strategy, CV_params_for_algo, CV_params);
toc
fprintf('CV parameters are now\n')
model.cv
model.refit
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
accuracy = length(find((ypred ~= Ytst))) / length(ypred) %accuracy
%regression example
load diabetes
algo_name = 'RandomForestRegressor';
Xtrn = diabetes.x;
Ytrn = diabetes.y;
Xtst = Xtrn;
Ytst = Ytrn;
fprintf('Tutorial 5: running %s with default parameters\n', algo_name);
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name);
toc
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
norm_diff = norm(ypred - Ytst)/sqrt(length(ypred)) %norm difference
%now we send in some algo_params, lets try with n_estimators=50
fprintf('Tutorial 6: running %s with additional parameter of num estimators = 50\n', algo_name);
%create a dict(ionary) object and pass in the argument. Make sure if its
%int or float or whatever type required in scikit algorithm
algo_params = py.dict;
update(algo_params, py.dict(pyargs('n_estimators',int64(50))))
update(algo_params, py.dict(pyargs('n_jobs',int64(num_threads))))
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name, algo_params);
toc
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
norm_diff = norm(ypred - Ytst)/sqrt(length(ypred)) %norm difference
%now we use some form of CV, lets try GridSearchCV, with search over
%lets try max_depth, learning_rate and gamma for RandomForest
fprintf('Tutorial 7: running %s with additional parameter of GridSearchCV and trying (max_features, n_estimators) search\n', algo_name);
algo_params = py.dict;
CV_strategy = 'GridSearchCV';
CV_params_for_algo = py.dict;
update(CV_params_for_algo, py.dict(pyargs('n_estimators', int32(100:200:500) )))
update(CV_params_for_algo, py.dict(pyargs('max_features', int32(1:2:10) )))
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name, algo_params, CV_strategy, CV_params_for_algo);
toc
fprintf('model parameter searched over\n')
model.param_grid
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
norm_diff = norm(ypred - Ytst)/sqrt(length(ypred)) %norm difference
%now we use some form of CV, lets try GridSearchCV, with search over
%lets try max_depth, learning_rate and gamma for RandomForest
%and pass parameters to GridSearchCV
fprintf('Tutorial 8: running %s with additional parameter of GridSearchCV and trying (max_features, n_estimators) search\n', algo_name);
fprintf('Furthermore we now select a CV time of 4 and do refit=true\n')
algo_params = py.dict;
CV_strategy = 'GridSearchCV';
CV_params_for_algo = py.dict;
update(CV_params_for_algo, py.dict(pyargs('n_estimators', int32(100:200:500) )))
update(CV_params_for_algo, py.dict(pyargs('max_features', int32(1:2:10) )))
%https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html
CV_params = py.dict;
update(CV_params, py.dict(pyargs('cv', int64(4) )))
update(CV_params, py.dict(pyargs('refit', int64(true) )))
update(CV_params, py.dict(pyargs('n_jobs', int64(-1) ))) %-1 means all processors
tic
model = scikit_train_supervised(Xtrn, Ytrn, algo_name, algo_params, CV_strategy, CV_params_for_algo, CV_params);
toc
fprintf('CV parameters are now: CV=%s, model.refit=%s\n', char(model.cv),char(model.refit))
tic
ypred = scikit_predict_supervised(Xtst, model);
toc
norm_diff = norm(ypred - Ytst)/sqrt(length(ypred)) %norm difference