- "update the client layout with more available parameters...

- "update the client layout with more available parameters cross-validation(`kfold`/`shufflesplit`/`StratifiedShuffleSplit`)
- support the `n_splits` as a slider element for use to adjust the folds
- update the `confusion matrix table`show button

.gitignore

 *.wav
 *.npy
+*.nyc
 *.csv
\ No newline at end of file

.vscode/settings.json

@@ -4,5 +4,9 @@
     "python.linting.pylintEnabled": true,
     "git.ignoreLimitWarning": true,
     "python.linting.lintOnSave": false,
-    "python.analysis.typeCheckingMode": "off"
+    "python.analysis.typeCheckingMode": "off",
+    "[python]": {
+        "editor.defaultFormatter": "ms-python.black-formatter"
+    },
+    "python.formatting.provider": "none"
 }
\ No newline at end of file

README.md

@@ -56,7 +56,6 @@
 #### 需要注意的包
 
 - librosa 
-
   - librosa 0.9.2 is not the lastes version,but the newest version don't work well with some matplotlib version
   - these problems appeared that I install `matplotlib` with `conda install` and `librosa`with `pip isntall`
   - The compatibility issues may be caused by mixing two installation methods.
@@ -64,15 +63,59 @@
   - However, the newer version may become the more preferred choice in the future, once the bugs or compatibility issues have been fixed.
 
 - pluggy
-
   - the pluggy may installed automatically or not(I write here becases it when I test the `requirements.txt` in a brand new conda environment ,the pip prompt me that the pluggy was not installed 
 
 - tensorflow
-
   - if you just want to experience the basic ML alogrithms' working on SER task,it's no need for you to install tensorflow
   - in may case , I use tensorflow==2.10,but other version of tensorflow above 2.6 may work well too
 
-  
+
+### 其他情况说明
+
+- 我在一次调试matplotlib backend设置的过程中,偶然发现在notebook中使用`%matplotlib qt`失败
+
+- 后来我尝试卸载重装matplotlib和pyside
+
+- 然而提示我一些`dll`问题和权限错误问题
+
+- 这些问题时平时不曾遇到
+
+- 我猜测可能时机器太久没有关机了(平时我都是休眠),导致系统出现了一些错误
+
+  - 机器发生错误是很有可能的,就比如学校图书馆的刷脸系统,验证通过平时屏幕显示绿色,然而最近通过显示的也是红色
+  - 而在早期的windows7上,有时候从休眠中回复直接会失败
+
+- 然后我重启机器,发现任务栏多出了个搜索框,系统更新了没有重启可能也造成了一些影响
+
+- 有时候还会出现各种意外的问题
+
+  - 例如`KNeighborsClassifier`训练完毕后,用它预测新的样本或在训练集上预测会报错:
+
+    - ```python
+      
+      File d:\condaPythonEnvs\tf210\lib\site-packages\sklearn\neighbors\_classification.py:237, in KNeighborsClassifier.predict(self, X)
+          235     neigh_dist = None
+          236 else:
+      --> 237     neigh_dist, neigh_ind = self.kneighbors(X)
+          239 classes_ = self.classes_
+          240 _y = self._y
+      
+          643 get_config = getattr(self._dynlib, "openblas_get_config",
+          644                      lambda: None)
+          645 get_config.restype = ctypes.c_char_p
+      --> 646 config = get_config().split()
+          647 if config[0] == b"OpenBLAS":
+          648     return config[1].decode("utf-8")
+      
+      AttributeError: 'NoneType' object has no attribute 'split'
+      ```
+
+    - 而当我切换到另一个环境又可以工作,说明是应该不是系统问题,而更可能是环境出了问题
+
+#### 意料之外的错误的解决方案
+
+- 重启你的机器,这或许解决一些系统错误
+- 创建/切换虚拟环境,也许是安装某些包引起依赖降级或者环境污染
 
 #### 关于进度条显示:tqdm
 
@@ -397,6 +440,104 @@ The initial letter(s) of the file name represents the emotion class, and the fol
 - BaggingClassifier
 - Recurrent Neural Networks (Keras)
 
+### 选择合适的分类器(用k-fold评估)
+
+- ```python
+  from sklearn.datasets import load_iris
+  from sklearn.linear_model import LinearRegression
+  from sklearn.model_selection import cross_val_score
+  
+  # 加载iris(鸢尾花)数据集
+  X, y = load_iris(return_X_y=True)
+  
+  # 使用线性回归模型进行交叉验证
+  model = LinearRegression()
+  
+  scores = cross_val_score(model, X, y, cv=5)
+  print("Scores:", scores)
+  print("Mean score:", scores.mean())
+  ```
+
+  - ```bash
+    Scores: [0.96666667 0.96666667 0.9        0.96666667 1.        ]
+    Mean score: 0.9600000000000002
+    ```
+
+- ```python
+  # 加载iris(鸢尾花)数据集
+  X, y = load_iris(return_X_y=True)
+  
+  # model=RandomForestClassifier()
+  model=SVC()
+  scores = cross_val_score(model, X, y, cv=5,verbose=3)
+  print("Scores:", scores)
+  print("Mean score:", scores.mean())
+  ```
+
+  - ```bash
+    Scores: [0.96666667 0.96666667 0.96666667 0.93333333 1.        ]
+    Mean score: 0.9666666666666666
+    ```
+
+- 而使用决策树模型进行分类,效果比较好,在5折叠验证中,都在0.9以上
+
+- 使用SVC或者RF,效果更加好
+
+#### 使用shuffle
+
+- 不对有序数据集iris洗牌,效果:
+
+  ```python
+  
+  # 加载iris(鸢尾花)数据集
+  X, y = load_iris(return_X_y=True)
+  
+  # 定义5折交叉验证
+  kf = KFold(
+      n_splits=5,
+      #    shuffle=True,
+      # random_state=42,
+  )
+  
+  # 使用线性回归模型进行训练和测试
+  model = LinearRegression()
+  # model=RandomForestClassifier()
+  scores_cv = []
+  for train_index, test_index in kf.split(X):
+      X_train, X_test = X[train_index], X[test_index]
+      y_train, y_test = y[train_index], y[test_index]
+      model.fit(X_train, y_train)
+      score = model.score(X_test, y_test)
+      scores_cv.append(score)
+      print("Score:", score)
+  mean_score = np.mean(scores_cv)
+  print(f"{mean_score=}")
+  ```
+
+  - ```bash
+    Score: 0.0
+    Score: 0.8512492308414581
+    Score: 0.0
+    Score: 0.7615543936085848
+    Score: 0.0
+    mean_score=0.32256072489000853
+    ```
+
+  - 通常出现这种情况的话,可以认为是数据集读入策略有问题
+
+- 解开shuffle=True的注释,得到正常的预测性能
+
+- ```bash
+  Score: 0.9468960016420045
+  Score: 0.9315787260143983
+  Score: 0.9177129838664249
+  Score: 0.9026578332122843
+  Score: 0.921073136533955
+  mean_score=0.9239837362538135
+  ```
+
+  
+
 #### SVC
 
 - Scikit-learn中的SVC是一种支持向量机（Support Vector Machine）分类器，用于解决二分类和多分类问题。SVC是一种非常强大的模型，可以处理高维度的数据，并且能够有效地处理非线性可分的数据。
@@ -414,7 +555,7 @@ The initial letter(s) of the file name represents the emotion class, and the fol
 - 在Scikit-learn中，SVC的使用非常简单，只需要创建一个SVC对象，设置一些超参数，然后调用fit()方法训练模型。可以使用predict()方法对新数据进行分类预测。
 - 总之，SVC是一种强大的分类器，适用于解决二分类和多分类问题，尤其擅长处理高维度和非线性可分的数据。
 
-#### sklearn.svm.svc
+##### sklearn.svm.svc
 
 - [`sklearn.svm`](https://scikit-learn.org/stable/modules/classes.html#module-sklearn.svm).SVC
 

SG/ccser_gui.py

 ##
-import os
 import inspect
+import os
+
 import constants.beauty as bt
+import constants.uiconfig as ufg
 import data_visualization as dv
 import fviewer
 import ipdb
+import numpy as np
 import PySimpleGUI as sg
 import query as q
-import constants.uiconfig as ufg
-import constants.beauty as bt
-from constants.beauty import (
-    ccser_theme,
-    db_introduction,
-    h2,
-    logo,
-    option_frame,
-    result_frame,
-)
-from demo_programs.Demo_Nice_Buttons import image_file_to_bytes, red_pill64, wcolor
+from constants.beauty import (ccser_theme, db_introduction, h2, logo,
+                              option_frame, result_frame)
+from constants.uiconfig import ML_KEY, __version__
+from demo_programs.Demo_Nice_Buttons import (image_file_to_bytes, red_pill64,
+                                             wcolor)
 from fviewer import audio_viewer_layout, fviewer_events, selected_files
 from joblib import load
 from multilanguage import get_your_language_translator
-from constants.uiconfig import ML_KEY, __version__
 from user import UserAuthenticatorGUI
 
+from config.algoparams import ava_cv_modes
 
 lang = get_your_language_translator("English")
 import sys
@@ -33,16 +30,9 @@ import sys
 from audio.core import get_used_keys
 from audio.graph import showFreqGraph, showMelFreqGraph, showWaveForm
 from config.EF import ava_algorithms, ava_emotions, ava_features
-from config.MetaPath import (
-    ava_dbs,
-    bclf,
-    brgr,
-    emodb,
-    get_example_audio_file,
-    ravdess,
-    savee,
-    speech_dbs_dir,
-)
+from config.MetaPath import (ava_dbs, bclf, brgr, emodb,
+                             get_example_audio_file, ravdess, savee,
+                             speech_dbs_dir)
 
 
 def import_config_bookmark():
@@ -58,7 +48,7 @@ def define_constants():
 
 size = (1500, 1000)
 # size=None
-
+# ava_cv_modes=("kfold","ss","sss")
 train = "trian"
 test = "test"
 algorithm = ""
@@ -66,7 +56,18 @@ audio_selected = ""
 speech_folder = speech_dbs_dir
 no_result_yet = f"No Result Yet"
 predict_res_key = "emotion_predict_res"
+kfold_radio_key = "kfold"
+shuffle_split_radio_key = "ss"
+show_confusion_matrix_key = "show_confusion_matrix"
+
+stratified_shuffle_split_radio_key = "sss"
+current_model_key = "current_model"
+current_model_tip_key = "current_model_tip"
 predict_proba_tips_key = "predict_proba"
+cv_splits_slider_key = "cv_splits_slider"
+
+train_cv_result_table_key = "train_cv_result_table"
+# test_score&train_score view
 train_result_table_key = "train_result_table"
 predict_proba_table_key = "predict_proba_table"
 predict_proba_frame_key = "predict_proba_frame"
@@ -76,7 +77,7 @@ userUI = UserAuthenticatorGUI()
 
 
 ##
-def get_algos_elements_list():
+def get_algos_elements_list(ava_algorithms=ava_algorithms):
     """
       Radio组件需要设置分组,这里分组就设置额为:algorithm
     利用default参数设置选中默认值(默认选中第一个)
@@ -98,7 +99,6 @@ def get_algos_elements_list():
                 enable_events=True,
             )
         )
-
     return algos_radios
 
 
@@ -238,8 +238,85 @@ def make_window(theme=None, size=None):
                 num_rows=1,  # 默认表格会有一定的高度,这里设置为1,避免出现空白
                 hide_vertical_scroll=True,
             )
+        ],
+        [
+            sg.Table(
+                values=[["pending"] * 2],
+                headings=["fold", "accu_score"],
+                justification="center",
+                font="Arial 16",
+                expand_x=True,
+                key=train_cv_result_table_key,
+                num_rows=1,  # 默认表格会有一定的高度,这里设置为1,避免出现空白
+                hide_vertical_scroll=True,
+                visible=False,
+            )
+        ],
+        # [
+        #     sg.Table(
+        #         values=[["pending"] * 2],
+        #         headings=["confusion_matrix", "accu_score"],
+        #         justification="center",
+        #         font="Arial 16",
+        #         expand_x=True,
+        #         key=train_cv_result_table_key,
+        #         num_rows=1,  # 默认表格会有一定的高度,这里设置为1,避免出现空白
+        #         hide_vertical_scroll=True,
+        #         visible=False,
+        #     )
+        # ]
+    ]
+    confution_matrix_button_layout=[
+        [sg.B("show confusion matrix", key=show_confusion_matrix_key)],
+    ]
+
+    cv_mode_layout = [
+        [
+            sg.T("cv mode:"),
+            sg.Radio(
+                "k-fold",
+                group_id="cv_mode",
+                key=kfold_radio_key,
+                default=False,
+                enable_events=True,
+            ),
+            sg.Radio(
+                "shuffle-split",
+                group_id="cv_mode",
+                key=shuffle_split_radio_key,
+                default=False,
+                enable_events=True,
+            ),
+            sg.Radio(
+                "stratified-shuffle-split",
+                group_id="cv_mode",
+                key=stratified_shuffle_split_radio_key,
+                default=True,
+                enable_events=True,
+            ),
         ]
     ]
+    cv_param_settings_layout = [
+        [
+            sg.T("cv splits:"),
+            sg.Slider(
+                range=(1, 10),
+                key=cv_splits_slider_key,
+                orientation="h",
+                expand_x=True,
+                default_value=5,
+                enable_events=True,
+            ),
+        ],
+        *cv_mode_layout,
+    ]
+    other_settings_frame_layout = [
+        [
+            bt.option_frame(
+                title="Other Parameter Settings", layout=cv_param_settings_layout
+            ),
+        ],
+    ]
     train_fit_button_layout = [
         [
             # sg.Button('start train'),
@@ -252,9 +329,12 @@ def make_window(theme=None, size=None):
                 pad=(0, 0),
                 key="start train",
             ),
+            sg.pin(sg.T("current model:", key=current_model_tip_key, visible=False)),
+            sg.T("", key=current_model_key),
         ]
     ]
-    train_result_layout = [
+
+    train_result_frame_layout = [
         [
             bt.result_frame(
                 title=lang["train_result_title"],
@@ -268,11 +348,11 @@ def make_window(theme=None, size=None):
         text=no_result_yet, justification="c", key=predict_res_key
     )
     # predict_proba_tips_layout = [[sg.Text("pending", key=predict_proba_tips_key)]]
-    #默认不显示predict_proba的不可用说明
+    # 默认不显示predict_proba的不可用说明
     predict_proba_tips_layout = bt.normal_content_layout(
-        text="pending", key=predict_proba_tips_key,visible=False
+        text="pending", key=predict_proba_tips_key, visible=False
     )
-    #默认显示predict_proba表格
+    # 默认显示predict_proba表格
     predict_proba_table_layout = [
         [
             sg.Table(
@@ -289,10 +369,7 @@ def make_window(theme=None, size=None):
             )
         ]
     ]
-    predict_prob_layout=[
-        *predict_proba_tips_layout,
-        *predict_proba_table_layout
-    ]
+    predict_prob_layout = [*predict_proba_tips_layout, *predict_proba_table_layout]
     predict_res_frames_layout = [
         [result_frame(layout=predict_res_layout)],
         [
@@ -380,8 +457,10 @@ def make_window(theme=None, size=None):
         + e_config_layout
         + f_config_layout
         + algos_layout
+        + other_settings_frame_layout
         + train_fit_button_layout
-        + train_result_layout
+        + train_result_frame_layout
+        + confution_matrix_button_layout
         + file_choose_layout
         + predict_res_frames_layout
         + draw_layout
@@ -527,7 +606,7 @@ def initial(values=None, verbose=1):
     train_db = values["train_db"]
     test_db = values["test_db"]
     e_config = scan_choosed_options(values)
-    algorithm = selected_algo(values)
+    algorithm = selected_radio_in(values)
     f_config = selected_features(values)
     if verbose >= 2:
         # print(train_db, test_db, e_config, algorithm, f_config)
@@ -551,14 +630,13 @@ def selected_features(values):
     return f_config
 
 
-def selected_algo(values):
-    global algorithm
-    for algo in ava_algorithms:
-        if values and values[algo]:
-            # 获取选中的算法名称(key)
-            algorithm = algo
+def selected_radio_in(values,ava_list=ava_algorithms):
+    # global algorithm
+    # res=""
+    for algo_name in ava_list:
+        if values and values[algo_name]:
             break
-    return algorithm
+    return algo_name
 
 
 def scan_choosed_options(values):
@@ -644,18 +722,18 @@ def recognize_auido(
 
         data = list(predict_proba.items())
         # print(data,"@{data}")
-        data = [[emo, proba] for emo, proba in data]
-        #关闭proba_tip的显示
+        data = [[emo, round(proba,bt.score_ndigits)] for emo, proba in data]
+        # 关闭proba_tip的显示
         window[predict_proba_tips_key].update(visible=False)
-        #更新proba表格内容
+        # 更新proba表格内容
         # window[predict_proba_tips_frame_key].update(visible=False)
         ppt = window[predict_proba_table_key]
         # inspect.getfullargspec(ppt.update)
         ppt.update(
             values=data,
             num_rows=4,
-            #    display_row_numbers=True
-            visible=True
+            #display_row_numbers=True
+            visible=True,
         )
         # window[]
         # window[predict_proba_table_frame_key].update(visible=True)
@@ -667,7 +745,7 @@ def recognize_auido(
             ),
             visible=True,
         )
-        #关闭表格的显示
+        # 关闭表格的显示
         window[predict_proba_table_key].update(visible=False)
     window.refresh()
 
@@ -696,11 +774,7 @@ def start_train_model(
     from recognizer.basic import EmotionRecognizer
 
     if verbose:
-        print("train_db:", train_db)
-        print("test_db:", test_db)
-        print("e_config:", e_config)
-        print("f_config:", f_config)
-        print("algorithm:", algorithm)
+        check_training_arguments(train_db, test_db, e_config, f_config, algorithm)
 
     bclf_estimators = load(bclf)
 
@@ -719,7 +793,6 @@ def start_train_model(
 
     if algorithm == "RNN":
         from recognizer.deep import DeepEmotionRecognizer
-
         der = DeepEmotionRecognizer(
             train_dbs=train_db, test_dbs=test_db, e_config=e_config, f_config=f_config
         )
@@ -752,6 +825,7 @@ def model_res(er, verbose=1):
     """
     train_score = er.train_score()
     test_score = er.test_score()
+
     if verbose:
         print(f"{er.model=}")
         print(f"{test_score=}")
@@ -776,7 +850,7 @@ def main(verbose=1):
 
     while True:
         if verbose >= 2:
-            check_training_arguments(e_config, f_config, train_db, test_db, algorithm)
+            check_training_arguments(train_db, test_db, e_config, f_config, algorithm)
 
         if event:  # 监听任何event
             print(event, "@{event}", __file__)
@@ -812,7 +886,7 @@ def main(verbose=1):
                 print(f_config, "@{f_config}")
 
         elif event in ava_algorithms:
-            algorithm = selected_algo(values)
+            algorithm = selected_radio_in(values)
             if verbose:
                 print(algorithm, "@{algorithm}")
                 # print(event, "处于选择algorithm的循环中.")
@@ -832,6 +906,7 @@ def main(verbose=1):
             # print("完成文件选取")
         # --情感识别阶段--
         elif event == "start train":
+            n_splits = values[cv_splits_slider_key]
             er = start_train_model(
                 train_db=train_db,
                 test_db=test_db,
@@ -839,15 +914,10 @@ def main(verbose=1):
                 f_config=f_config,
                 algorithm=algorithm,
             )
+
             # 训练收尾工作:将计算结果(识别器)传递给fviewer,赋能fviewer可以(直接利用识别器对象)进行识别
 
-            fviewer.er = er  # 是否为多余#TODO
-            train_score, test_score = model_res(er, verbose=verbose)
-            # window["train_result"].update(f"{train_score=},{test_score=}")
-            res = [round(x, 4) for x in (train_score, test_score)]
-            window["train_result_table"].update(
-                values=[res]
-            )  # values类型是list[list[any]],每个内部列表表示表格的一个行的数据
+            refresh_trained_view(verbose, window, er, values)
 
         elif event == "recognize it":
             recognize_auido(
@@ -891,6 +961,11 @@ def main(verbose=1):
             content = [logo, db_introduction]
             res = "\n".join(content)
             sg.popup_scrolled(res, size=(150, 100), title="Introduction")
+        elif event==show_confusion_matrix_key:
+                from SG.demo_pandas_table import TablePandas
+                cm=er.confusion_matrix()
+                tp=TablePandas(df=cm)
+                tp.show_confution_matrix_window()
         else:
             # 具有独立的事件循环,直接调用即可
             userUI.run_module(event, values, window=window, verbose=1)
@@ -908,6 +983,47 @@ def main(verbose=1):
     window.close()
 
 
+def refresh_trained_view(verbose, window, er, values):
+    """
+    Refreshes the trained view with the given parameters.
+    these args are available for the table element to update
+    args=ArgSpec(args=['self', 'values', 'num_rows', 'visible', 'select_rows', 'alternating_row_color', 'row_colors'],
+
+    varargs=None, keywords=None, defaults=(None, None, None, None, None, None))🎈
+    Args:
+    verbose (bool): Whether to print verbose output or not.
+    window (sg.Window): The PySimpleGUI window object to update.
+    er (EvaluationResult): The evaluation result object to use for updating the view.
+    """
+    fviewer.er = er  # 是否为多余#TODO
+    train_score, test_score = model_res(er, verbose=verbose)
+    # window["train_result"].update(f"{train_score=},{test_score=}")
+    res = [round(x, bt.score_ndigits) for x in (train_score, test_score)]
+    window[train_result_table_key].update(
+        values=[res]
+    )  # values类型是list[list[any]],每个内部列表表示表格的一个行的数据
+    window[current_model_tip_key].update(visible=True)
+    window[current_model_key].update(value=er.model)
+    n_splits = values[cv_splits_slider_key]
+    # cv_mode=values[kfold_radio_key]
+    cv_mode=selected_radio_in(values,ava_list=ava_cv_modes)
+    # print(cv_mode,"@{cv_mode}🎈")
+
+    fold_scores = er.model_cv_score(mean_only=False, n_splits=n_splits,cv_mode=cv_mode)
+    folds = len(fold_scores)
+    mean_score = np.mean(fold_scores)
+    fold_scores_rows = [
+        [str(f"{i+1}"), round(score, bt.score_ndigits)]
+        for i, score in enumerate(fold_scores)
+    ]
+    fold_scores_rows.append(["mean_score", round(mean_score, bt.score_ndigits)])
+    tcrt = window[train_cv_result_table_key].update
+    # args=inspect.signature(tcrt)
+    # print(f"{args=}🎈")
+    tcrt(values=fold_scores_rows, num_rows=folds + 1, visible=True)
+
+
+
 def open_folder_event(window):
     print("[LOG] Clicked Open Folder!")
     folder_or_file = sg.popup_get_folder(
@@ -946,7 +1062,7 @@ def file_selected_record(verbose, event, values):
         print(event, values["-FILENAME-"])
 
 
-def check_training_arguments(e_config, f_config, train_db, test_db, algorithm):
+def check_training_arguments(train_db, test_db, e_config, f_config, algorithm):
     print(f"train_db = {train_db}")
     print(f"test_db = {test_db}")
     print(f"e_config = {e_config}")

SG/constants/beauty.py

@@ -12,6 +12,7 @@ frame_size = (600, 50)
 lb_size = (60, 10)
 ml_size = (60, 20)
 seperator_color = "blue"
+score_ndigits=4
 welcom_title_size = (45, 1)
 slider_size = (60, 10)
 ccser_theme = "Reddit"
@@ -329,6 +330,7 @@ if __name__ == "__main__":
     # layout_inner = [[sg.Text("demo")]]
     # layout = [[result_frame(title="demo", layout=layout_inner)]]
     # layout=res_content_layout("demo", expand_x=True)
+    
     layout=normal_content_layout("demo")
     window = sg.Window("demo of beauty elements", layout,resizable=True)
     window.read()

SG/demo_pandas_table.py

+import PySimpleGUI as sg
+import pandas as pd
+import SG.constants.beauty as bt
+
+sg.theme(bt.ccser_theme)
+class TablePandas():
+    def __init__(self,df=None) -> None:
+        if df is None:
+            # 创建一个 Pandas 数据帧
+            demo_data = {
+                "Name": ["Alice", "Bob", "Charlie", "David"],
+                "Age": [25, 30, 35, 40],
+                "Salary": [50000, 60000, 70000, 80000],
+            }
+
+            df = pd.DataFrame(demo_data)
+        self.df=df
+    
+    # 创建 PySimpleGUI 窗口布局
+    def create_table_window(self,df):
+        layout = [
+            [
+                sg.Table(
+                    values=df.values.tolist(),
+                    headings=df.columns.tolist(),
+                    max_col_width=25,
+                    auto_size_columns=True,
+                    justification="center",
+                    num_rows=min(25, len(df)),
+                )
+            ],
+            # [sg.Button("Exit")],
+        ]
+
+        return layout
+    def get_confution_matrix_window(self,df=None):
+        layout = self.create_table_window(df)
+        window = sg.Window("Pandas Table Viewer", layout)
+        return window
+    # def show_confution_matrix_window(df):
+    #     window=get_confution_matrix_window(df)
+        
+
+    def show_confution_matrix_window(self,df=None):
+        # 创建 PySimpleGUI 窗口
+        # window = sg.Window("Pandas Table Viewer", layout)
+        df=df if df else self.df
+        window=self.get_confution_matrix_window(df=df)
+        # 处理事件循环
+        while True:
+            event, values = window.read()
+            if event in (sg.WINDOW_CLOSED,):
+                break
+
+        # 关闭 PySimpleGUI 窗口
+        window.close()
+if __name__=="__main__":
+    tp=TablePandas()
+    tp.show_confution_matrix_window()
+    # show_confution_matrix_window(df)
\ No newline at end of file

SG/fviewer.py

@@ -445,7 +445,7 @@ def fviewer_events(window, event=None, values=None, verbose=1):
             print(emo_res, "@{emo_res}")
             print(abs_pathes, "@{abs_pathes}")
 
-            t = ts.TableShow(header=["emotion", "path"], lists=[emo_res, abs_pathes])
+            t = ts.TableShow(header=["emotion", "path"], data_lists=[emo_res, abs_pathes])
             print(t.lists, "@{t.lists}")
             t.run()
 

SG/h.py

-import humansize
-
-size = 1073741824  # 1 GB 的字节数
-
-# 将字节数转换为人类可读的格式
-print(humansize.approximate_size(size, binary=True))  # 输出：1.0 GiB
-print(humansize.approximate_size(size, binary=False))  # 输出：1.1 GB

SG/q.md

-*   Python GUIs for Humans[PySimpleGUI User's Manual](https://www.pysimplegui.org/en/latest/#pysimplegui-users-manual)[Jump-Start](https://www.pysimplegui.org/en/latest/#jump-start)[2021 Updates....](https://www.pysimplegui.org/en/latest/#2021-updates)[About The PySimpleGUI Documentation System](https://www.pysimplegui.org/en/latest/#about-the-pysimplegui-documentation-system)[Platforms](https://www.pysimplegui.org/en/latest/#platforms)[The PySimpleGUI "Family"](https://www.pysimplegui.org/en/latest/#the-pysimplegui-family)[Support](https://www.pysimplegui.org/en/latest/#support)[Learning Resources](https://www.pysimplegui.org/en/latest/#learning-resources)[The Quick Tour](https://www.pysimplegui.org/en/latest/#the-quick-tour)[Some Examples](https://www.pysimplegui.org/en/latest/#some-examples)[Pi Windows](https://www.pysimplegui.org/en/latest/#pi-windows)[Games](https://www.pysimplegui.org/en/latest/#games)[Windows Programs That Look Like Windows Programs](https://www.pysimplegui.org/en/latest/#windows-programs-that-look-like-windows-programs)[Background - Why PySimpleGUI Came to Be](https://www.pysimplegui.org/en/latest/#background-why-pysimplegui-came-to-be)[Features](https://www.pysimplegui.org/en/latest/#features)[Getting Started with PySimpleGUI](https://www.pysimplegui.org/en/latest/#getting-started-with-pysimplegui)[PEP8 Bindings For Methods and Functions](https://www.pysimplegui.org/en/latest/#pep8-bindings-for-methods-and-functions)[High Level API Calls - Popup's](https://www.pysimplegui.org/en/latest/#high-level-api-calls-popups)[Progress Meters!](https://www.pysimplegui.org/en/latest/#progress-meters)[Debug Output (easy\_print = Print = eprint)](https://www.pysimplegui.org/en/latest/#debug-output-easy_print-print-eprint)[Custom window API Calls (Your First window)](https://www.pysimplegui.org/en/latest/#custom-window-api-calls-your-first-window)[Copy these design patterns!](https://www.pysimplegui.org/en/latest/#copy-these-design-patterns)[Building Custom Windows](https://www.pysimplegui.org/en/latest/#building-custom-windows)[Themes - Automatic Coloring of Your Windows](https://www.pysimplegui.org/en/latest/#themes-automatic-coloring-of-your-windows)[Window Object - Beginning a window](https://www.pysimplegui.org/en/latest/#window-object-beginning-a-window)[Layouts](https://www.pysimplegui.org/en/latest/#layouts)[Generated Layouts (For sure want to read if you have > 5 repeating elements/rows)](https://www.pysimplegui.org/en/latest/#generated-layouts-for-sure-want-to-read-if-you-have-5-repeating-elementsrows)[Elements](https://www.pysimplegui.org/en/latest/#elements)[SystemTray](https://www.pysimplegui.org/en/latest/#systemtray)[Global Settings](https://www.pysimplegui.org/en/latest/#global-settings)[Persistent windows (Window stays open after button click)](https://www.pysimplegui.org/en/latest/#persistent-windows-window-stays-open-after-button-click)[Updating Elements (changing element's values in an active window)](https://www.pysimplegui.org/en/latest/#updating-elements-changing-elements-values-in-an-active-window)[Cursors - Setting for Elements and Windows](https://www.pysimplegui.org/en/latest/#cursors-setting-for-elements-and-windows)[Keyboard & Mouse Capture](https://www.pysimplegui.org/en/latest/#keyboard-mouse-capture)[Menus](https://www.pysimplegui.org/en/latest/#menus)[TTK & TTK Scrollbars](https://www.pysimplegui.org/en/latest/#ttk-ttk-scrollbars)[Running Multiple Windows](https://www.pysimplegui.org/en/latest/#running-multiple-windows)[The PySimpleGUI Debugger](https://www.pysimplegui.org/en/latest/#the-pysimplegui-debugger)[User Settings API](https://www.pysimplegui.org/en/latest/#user-settings-api)[Timer API](https://www.pysimplegui.org/en/latest/#timer-api)[Extending PySimpleGUI](https://www.pysimplegui.org/en/latest/#extending-pysimplegui)[Troubleshooting](https://www.pysimplegui.org/en/latest/#troubleshooting)[Debug Output](https://www.pysimplegui.org/en/latest/#debug-output)["Demo Programs" Applications](https://www.pysimplegui.org/en/latest/#demo-programs-applications)[Creating a Windows .EXE File](https://www.pysimplegui.org/en/latest/#creating-a-windows-exe-file)[Creating a Mac App File](https://www.pysimplegui.org/en/latest/#creating-a-mac-app-file)[Known Issues](https://www.pysimplegui.org/en/latest/#known-issues)
\ No newline at end of file

SG/table_show.py

@@ -2,7 +2,7 @@ import PySimpleGUI as sg
 import pandas as pd
 from config.MetaPath import recognize_result_dir
 class TableShow():
-    def __init__(self,header=None,lists=None):
+    def __init__(self,header=None,data_lists=None):
         """将二维列表作为表格数据显示
 
         Parameters
@@ -11,14 +11,15 @@ class TableShow():
             _description_
         """
 
-        self.lists=lists
-        self.length=len(lists[0])
+        self.lists=data_lists
+        self.length=len(data_lists[0])
         # 创建表格数据
-        self.data_rows = [[l[i] for l in lists] for i in range(self.length)]
+        self.data_rows = [[l[i] for l in data_lists] for i in range(self.length)]
         # print(self.data_rows,"@{data}")
         # 定义表头
         # header = ["c1","c2"]
         self.header=header#columns
+
         self.data_df=pd.DataFrame(self.data_rows,columns=self.header)
         # 创建表格布局
         warning="the save operation will comsume  some time to complete!Be patient!"
@@ -42,7 +43,7 @@ class TableShow():
 
         ]
     def run(self):
-        window = sg.Window("结果表格", self.layout,resizable=True,size=(500,400))
+        window = sg.Window("result table", self.layout,resizable=True,size=(500,400))
         # 事件循环
         while True:
             event, values = window.read()
@@ -73,6 +74,6 @@ class TableShow():
 if __name__=="__main__":
         
     # 创建窗口
-    ts=TableShow(header=list("ABC"),lists=[[1,2,2],[3,4,4],[5,6,6]])
+    ts=TableShow(header=list("ABC"),data_lists=[[1,2,2],[3,4,4],[5,6,6]])
     ts.run()
     

SG/tmp.py

-# # a=1
-# # b=a
-# # print(f"{a=} {b=}; {id(a)=} {id(b)=}")
-# # a=100
-# # print(f"{a=} {b=}; {id(a)=} {id(b)=}")
-# # test=100
-# # print(f"{test=} {id(test)=}")
-
-# ##
-# print(f"{id(10)=}")
-# a=10
-# print(f"{id(a)=}")
-
-# def f(x):
-#     print(f"{id(x)=}")
-#     x=20
-#     print(f"{id(20)=}")
-#     print(f"{id(x)=}")
-
-# f(a)
-# a
-# print('a: ', a)#10
-a=[1,2,3]
-def f(x):
-	x=100
-f(a)
-print(f"{a=}")#[1,2,3]
-
-b=[1,2,3]
-def g(x):
-	x[0]=100
-g(b)
-print(f"{b=}")
-

SG/tmp2

-import PySimpleGUI as sg
-
-# 创建窗口和 Frame 元素
-layout = [
-    [sg.Frame('Frame 1', [[sg.Text('Frame content')]], key='-FRAME-')],
-    [sg.Button('Update Frame')],
-]
-window = sg.Window('My Window', layout)
-
-# 事件循环
-while True:
-    event, values = window.read()
-    if event == sg.WIN_CLOSED:
-        break
-    if event == 'Update Frame':
-        # 更新 Frame 布局
-        new_layout = [[sg.Text('Updated content')]]
-        
-        window['-FRAME-'].update(new_layout)
-
-# 关闭窗口
-window.close()
\ No newline at end of file

config/algoparams.py

+ava_cv_modes=("kfold","ss","sss")
\ No newline at end of file

grid/test.py

@@ -53,21 +53,16 @@ gb_best=(GradientBoostingClassifier(learning_rate=0.3, loss='log_loss', max_dept
 # bclf[1]=rf_best
 # bclf[2]=gb_best
 ##
-# dump(rf_best, "qq.joblib")
 # bclf[-1] = bag_best
 ##
 bclf_res=load("bclf.joblib")
 for item in bclf_res:
     print(item)
 
-# ##
-# brgr
-# ##
 # # 修改回归模型
 # bag_rgr_best = (BaggingRegressor(max_features=1, max_samples=0.1),
 #                 {'max_features': 1, 'max_samples': 0.1, 'n_estimators': 10},
 #                 0.6521001743540973)
 
-                
 # # brgr[-1]=bag_rgr_best
 

recognizer/basic.py

@@ -2,7 +2,7 @@
 # from typing_extensions import deprecated
 import random
 from time import time
-
+from config.algoparams import ava_cv_modes
 import matplotlib.pyplot as pl
 import numpy as np
 import pandas as pd
@@ -11,7 +11,7 @@ from sklearn.ensemble import RandomForestClassifier
 from sklearn.metrics import (accuracy_score, classification_report,
                              confusion_matrix, fbeta_score, make_scorer,
                              mean_absolute_error, mean_squared_error)
-from sklearn.model_selection import GridSearchCV
+from sklearn.model_selection import GridSearchCV, KFold, ShuffleSplit, StratifiedShuffleSplit, cross_val_score
 from sklearn.neighbors import KNeighborsClassifier
 from sklearn.svm import SVC
 from tqdm import tqdm
@@ -401,12 +401,14 @@ class EmotionRecognizer:
         y_test = self.y_test
         # 调用训练好的模型进行预测
         model = self.model if self.model is not None else self.best_model()
-        if len(X_test) == 0:
-            raise ValueError("X_test is empty")
-        if len(y_test) == 0:
-            raise ValueError("y_test is empty")
+        self.validate_empty_array(X_test=X_test,y_test=y_test)
+        # if len(X_test) == 0:
+        #     raise ValueError("X_test is empty")
+        # if len(y_test) == 0:
+        #     raise ValueError("y_test is empty")
         # 预测计算
-        print(X_test.shape, y_test.shape,"🎈")
+        if verbose:
+            print(X_test.shape, y_test.shape,"🎈")
         y_pred = model.predict(X_test)  # type: ignore
         if choosing == False:
             self.y_pred = np.array(y_pred)
@@ -419,7 +421,53 @@ class EmotionRecognizer:
             report = classification_report(y_true=y_test, y_pred=y_pred)
             print(report, self.model.__class__.__name__)
         return res
+    def model_cv_score(self, choosing=False, verbose=1,mean_only=True,n_splits=5,test_size=0.2,cv_mode="sss"):
+        """
+        使用交叉验证的方式来评估模型
+        Calculates score on testing data
+        """
+        X_train = self.X_train
+        y_train = self.y_train
+        # 调用训练好的模型进行预测
+        model = self.model if self.model is not None else self.best_model()
+        self.validate_empty_array(X_train, y_train)
+
+        # 预测计算
+        if verbose:
+            print(X_train.shape, y_train.shape,"🎈")
+            print(f"{n_splits=}")
+        n_splits=int(n_splits)
 
+        y_pred = model.predict(X_train)  # type: ignore
+        if choosing == False:
+            self.y_pred = np.array(y_pred)
+        # 交叉验证的方式评估模型的得分
+        cv_mode_dict=dict(
+            sss=StratifiedShuffleSplit(n_splits=n_splits, test_size=test_size, random_state=0),
+            ss=ShuffleSplit(n_splits=n_splits, test_size=test_size, random_state=0),
+            kfold=KFold(n_splits=n_splits, shuffle=True, random_state=0),
+        )
+        cv_mode_selected=cv_mode_dict[cv_mode]
+        if verbose:
+            print(f"{cv_mode=}🎈")
+        if self.classification_task:
+            # res = accuracy_score(y_true=y_test, y_pred=y_pred)
+            res=cross_val_score(model, X_train, y_train, cv=cv_mode_selected)
+            if mean_only:
+                res=res.mean()
+            
+        else:
+            res = mean_squared_error(y_true=y_train, y_pred=y_pred)
+        if self.verbose >= 2 or verbose >= 1:
+            report = classification_report(y_true=y_train, y_pred=y_pred)
+            print(report, self.model.__class__.__name__)
+        return res
+
+    def validate_empty_array(self, X_test=[], y_test=[]):
+        if len(X_test) == 0:
+            raise ValueError("X is empty")
+        if len(y_test) == 0:
+            raise ValueError("y is empty")
     def meta_paths_of_db(self, db, partition="test"):
         res = meta_paths_of_db(
             db=db,
@@ -528,12 +576,12 @@ class EmotionRecognizer:
             # make it percentage
             matrix *= 100
         if labeled:
-            matrix = pd.DataFrame(
+            matrix_df = pd.DataFrame(
                 matrix,
                 index=[f"true_{e}" for e in self.e_config],
                 columns=[f"predicted_{e}" for e in self.e_config],
             )
-        return matrix
+        return matrix_df
 
     def draw_confusion_matrix(self):
         """Calculates the confusion matrix and shows it"""
@@ -624,7 +672,7 @@ def main(EmotionRecognizer, e_config):
     # my_model = RandomForestClassifier(max_depth=3, max_features=0.2)
     my_model = SVC(C=0.001, gamma=0.001, kernel="poly",probability=True)
     my_model=KNeighborsClassifier(n_neighbors=3, p=1, weights='distance')
-    my_model = None
+    # my_model = None
 
     # rec = EmotionRecognizer(model=my_model,e_config=AHNPS,f_config=f_config_def,test_dbs=[ravdess],train_dbs=[ravdess], verbose=1)
     # rec = EmotionRecognizer(model=my_model,e_config=AHNPS,f_config=f_config_def,test_dbs=emodb,train_dbs=emodb, verbose=1)
@@ -646,6 +694,8 @@ def main(EmotionRecognizer, e_config):
     print(f"{train_score=}")
     test_score = er.test_score()
     print(f"{test_score=}")
+    cv_score=er.model_cv_score()
+    print(f"{cv_score=}")
 
     return er
 

recognizer/kfold_demo.py

+##
+import numpy as np
+from sklearn.datasets import load_iris
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.linear_model import LinearRegression
+from sklearn.model_selection import KFold, cross_val_predict, cross_val_score
+from sklearn.svm import SVC
+from sklearn.tree import DecisionTreeClassifier
+from sklearn.datasets import load_iris
+from sklearn.model_selection import ShuffleSplit, cross_val_score
+from sklearn.tree import DecisionTreeClassifier
+
+##
+
+# 加载iris(鸢尾花)数据集
+X, y = load_iris(return_X_y=True)
+
+#! 定义5折交叉验证
+kf = KFold(
+    n_splits=5,
+    shuffle=True,
+    random_state=42,
+)
+
+# 使用线性回归模型进行训练和测试
+model = LinearRegression()
+# model=RandomForestClassifier()
+
+scores_cv = []
+# 这里split参数可以是X也可以是y,因为只需要划分样本的索引,所以两者都可以
+for train_index, test_index in kf.split(y):
+    X_train, X_test = X[train_index], X[test_index]
+    y_train, y_test = y[train_index], y[test_index]
+    model.fit(X_train, y_train)
+    score = model.score(X_test, y_test)
+    scores_cv.append(score)
+    print("Score:", score)
+mean_score = np.mean(scores_cv)
+print(f"{mean_score=}")
+##
+#!使用cross_val_score
+#构造cv器的时候不需要传入数据集
+ss_cv = ShuffleSplit(n_splits=3, test_size=0.2, random_state=42)
+kf_cv=KFold(n_splits=3,shuffle=True,random_state=42)
+scores = cross_val_score(
+    model,
+    X,
+    y,
+    #cv=5,
+    #cv=ss_cv,
+    cv=kf_cv,
+    verbose=3,
+)
+#cv取整数时,采用的非随机化的kfold方法划分,不是很可靠
+#cv建议选用随机化的(StratifiedShuffleSplit最为高级)
+#cv取kfold对象时,我们可以选择shuffle=True,使得所有样本都能够参与训练集/测试集
+print("Scores:", scores)
+print("Mean score:", scores.mean())
+
+
+##
+
+# 使用决策树模型进行交叉验证，并对数据集进行随机化操作
+model = DecisionTreeClassifier()
+ss_cv = ShuffleSplit(n_splits=3, test_size=0.2, random_state=42)
+
+print("cv: ", ss_cv)
+
+# ssr=ss_cv.split(X,y)
+# for train_index,test_index in ssr:
+#     train_index,test_index=np.array(train_index),np.array(test_index)
+#     print(train_index.shape,test_index.shape)
+
+scores = cross_val_score(model, X, y, cv=ss_cv, verbose=True)
+print("Scores:", scores)
+print("Mean score:", scores.mean())
+
+
+##
+
+import numpy as np
+from sklearn.model_selection import StratifiedShuffleSplit
+
+# 使用随机生成数据测试
+rng = np.random.default_rng()
+rng.integers(20, size=(12, 2))
+# 样本总数为12,二分类,标签为0,1,两种样本比例为1:2
+n = 12
+n0, n1 = 1 * n // 3, 2 * n // 3
+# 随机的为这些模拟样本分配标签(因为这里不涉及到训练,所以随机分配标签不影响效果,在数据集划分的阶段,不用关心样本和标签的关联规律,如果是要训练,通常是不能随机给样本特征分配标签)
+
+y = [0] * n0 + [1] * n1
+y = np.array(y)
+rng.shuffle(y)
+# 下面一种方式采用概率的方式生成标签,但是即使样本总数n可以被3整除,生成的数组也不保证数量是1:2
+# y = rng.choice([0, 1], size=12, replace=True, p=[1/3, 2/3])
+# count=np.unique(y,return_counts=True)
+# print(count)
+# 为例放便验证,这里将标签数组和样本索引打印出来
+print(np.vstack([y, range(n)]))
+# 构造分层随机拆分对象,指定做独立的5次划分,每次划分,测试集的样本数量占总样本数量n的20%
+# 而StratifiedShuffleSplit会保持各个类别在测试集和训练集上的比例
+# 是两种独立的约束(例如0类样和1类样本比例在数据集中为1:2,那么在训练集和测试集中依然保持(或接近)1:2)
+sss = StratifiedShuffleSplit(n_splits=5, test_size=0.2, random_state=0)
+# print(sss)
+# 打印这5次
+for i, (train_index, test_index) in enumerate(sss.split(X, y)):
+    print(f"Fold {i}:")
+    print(f"  Train: index={train_index}")
+    print(f"  Test:  index={test_index}")
+    print(np.vstack((test_index, y[test_index])))
+##
+X,y=load_iris(return_X_y=True)
+#效果等同于
+data=load_iris()
+X=data.data
+y=data.target
+#由于Bunch对象的特性,可以用字典方式访问
+X=data['data']
+y=data['target']
+# 
+# 导出为pandas dataframe:
+frame_data=load_iris(as_frame=True)
+X_df=frame_data.data
+y_df=frame_data.target

release_note.md

@@ -5,8 +5,18 @@
 - for convenient,I tag the version of the project (mainly on the client ) with data time
 - the time(version) was create by powershell :
 
-  - ` Get-Date -Format "yyyy-MM-dd@HH:mm:ss"`
-
+  - ```powershell
+    Get-Date -Format "yyyy-MM-dd@HH:mm:ss"
+    ```
+  
+  - copy to clipbord automatically
+  
+    ```bash
+    Get-Date -Format "yyyy-MM-dd@HH:mm:ss"|scb
+    ```
+  
+    
+  
   
 
 ### versions with time
@@ -19,4 +29,12 @@
     - use the table element to show "test_score","train_score" and "predict_proba" if it is usable
     - update the file viwer part function and operation logic with more flexibility.
   - kwon problems:
-    - with the display or not display the predict_proba layout,the left column may need to resize the window to make the last part of view to visible!
+    - with the display or not display the predict_proba layout,the left column may need to resize the window to make the last part of view to visible!
\ No newline at end of file
+
+- 2023-04-27@22:04:52
+  - "update the client layout with more available parameters cross-validation(`kfold`/`shufflesplit`/`StratifiedShuffleSplit`)
+  - support the `n_splits` as a slider element for use to adjust th
+  - update the `confusion matrix table`show button 
+
+
+

requirements.txt

@@ -21,5 +21,6 @@ ipdb
 pydub
 pluggy
 psgdemos #this contain psgdemos and PySimpleGUI framework
+black #the python code formatter of th project
 # playsound
 # tensorflow==2.10.0
\ No newline at end of file