我在Chaquopy中利用joblib来load一个在pycharm中写好的AdaBoost模型：
from joblib import dump, load
filename = join(dirname(__file__), 'gao1_to_gao2_1.pkl')
clf = load(filename)
报错如下：
E/AndroidRuntime: FATAL EXCEPTION: main
Process: com.example.NCEPU, PID: 18024
com.chaquo.python.PyException: ValueError: Buffer dtype mismatch, expected 'SIZE_t' but got 'long long'
at .sklearn.tree._tree.Tree.__cinit__(_tree.pyx:607)
at .pickle.load_reduce(pickle.py:1587)
at .pickle.load(pickle.py:1210)
at .joblib.numpy_pickle._unpickle(numpy_pickle.py:504)
at .joblib.numpy_pickle.load(numpy_pickle.py:585)
at .AdaBoost.test(AdaBoost.py:76)
at .chaquopy_java.call(chaquopy_java.pyx:380)
at .chaquopy_java.Java_com_chaquo_python_PyObject_callAttrThrowsNative(chaquopy_java.pyx:352)
at com.chaquo.python.PyObject.callAttrThrowsNative(Native Method)
at com.chaquo.python.PyObject.callAttrThrows(PyObject.java:232)
at com.chaquo.python.PyObject.callAttr(PyObject.java:221)
at com.example.NCEPU.Student.Predict.AdaBoostActivity.callPythonCode(AdaBoostActivity.java:45)
at com.example.NCEPU.Student.Predict.AdaBoostActivity.createList(AdaBoostActivity.java:61)
at com.example.NCEPU.Student.Predict.AdaBoostActivity.lambda$initViews$1$AdaBoostActivity(AdaBoostActivity.java:56)
at com.example.NCEPU.Student.Predict.-$$Lambda$AdaBoostActivity$dJAvP4dDDrPleKcJgBCESW0ea0M.onClick(Unknown Source:2)
at android.view.View.performClick(View.java:7192)
at android.view.View.performClickInternal(View.java:7166)
at android.view.View.access$3500(View.java:824)
at android.view.View$PerformClick.run(View.java:27592)
at android.os.Handler.handleCallback(Handler.java:888)
at android.os.Handler.dispatchMessage(Handler.java:100)
at android.os.Looper.loop(Looper.java:213)
at android.app.ActivityThread.main(ActivityThread.java:8169)
at java.lang.reflect.Method.invoke(Native Method)
at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:513)
at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:1101)
翻遍stackoverflow发现是因为32位Python不能导入64位训练的模型，于是在Chaquopy中输出一下：
import platform
print(platform.architecture())
输出为：
('32bit', 'WindowsPE')
果然，Chaquopy中Python默认是32位。那没办法，只能用32位Python训练模型然后再导入了。
1.安装32位python枣庄人流医院哪个好 http://mobile.3635555.cn/
在官网下载32位的python38，然后安装，这个不再叙述。
安装后在Pycharm上配置：
这个时候在pycharm上打印：
import platform
print(platform.architecture())
输出为：
('32bit', 'WindowsPE')
可以看到当前版本已经是32位了
2.安装skleanr和pandas
直接在pycharm的Terminal中安装：
pip install sklearn -i https://pypi.douban.com/simple
pip install pandas -i https://pypi.douban.com/simple
安装过程有极大概率会报错，这中间省略了很多细节。。。。因为问题实在太多了，忘记记录了。不过最后是升级了一下pip就OK了。
3.训练模型并保存
这一步是在pycharm中进行的：
clf = AdaBoostClassifier(base_estimator=DecisionTreeClassifier(max_depth=7, min_samples_leaf=5),
n_estimators=200,
algorithm='SAMME', learning_rate=0.5)
clf = clf.fit(train_x, train_y) # 训练模型
dump(clf, 'gao1_to_gao2_1.pkl')
4.Chaquopy中加载模型
gao1_to_gao2_1.pkl放在python文件夹下：
from joblib import dump, load
filename = join(dirname(__file__), 'gao1_to_gao2_1.pkl')
clf = load(filename)