关联规则挖掘系统

这学期的大数据课设由三部分组成，记录自己负责的关联规则挖掘系统部分内容。

环境：代码由python实现，版本为3.7版本，界面设计采用tkinter，采用两种算法：Apriori和fpgrowth

Apriori算法使用python的模块efficient-apriori，

fpgrowth算法采用orange-associate模块，其实fpgrowth算法在python中有更好地封装：pyfpgrowth

系统的主界面如下图：

在做课设的时候，有一个要求的画出fp树，查阅网上资料后发现并没有好的实现方法，至少我试过了没有得到理想的效果

但是在github上找到一个html+js编写的画fptree的网页,点此跳转

于是就想着将这个网页嵌入python中，如果是pyqt实现的页面的话是比较好解决的，但是tkinter的话，最后找到唯一的一个方法就是使用cefpython3模块，可以内嵌浏览器。

此系统读入数据可放入txt文件中，格式如下:


数据可以是汉字，字母，数字。

本系统所有代码如下：

# -*- coding: utf-8 -*-
"""
Created on Thu Dec 19 22:15:06 2019

@author: shanhainanhua
"""

from tkinter import *
from tkinter import filedialog
#from fptree import *
#from pymining import itemmining,assocrules
from orangecontrib.associate.fpgrowth import *
from efficient_apriori import apriori
from orangecontrib.associate.fpgrowth import *
import functools


global minsupport
global minconfidence

data=[]
#事件函数


def LoadData():
    global data
    data=[]
    file_path=filedialog.askopenfilename(title='打开单个文件',
    filetypes=[('All Files', '*'),("文本文件", '*.txt')], # 只处理的文件类型
    initialdir='g:/') # 初始目录
    print(file_path)
    e1_value=file_path
    #e1.delete(1,END)
    e1.insert(END,e1_value)
    e1.grid(row=0, column=1,pady=18,padx=10)
    f=open(file_path,'rb').readlines()
   
    for line in f:
        #line = line.decode('gbk').strip()
        line = line.decode('UTF-8').strip()
        tmp=[]
        #line=line.split(",")
        line=line.split(" ")
        for item in line:
            tmp.append(str(item))
        data.append(tmp)
    print("输入的数据为{}".format(data))

   
def Start():
    #先获取组件的值
    global minsupport
    global minconfidence
    minsupport=float(e2.get())
    minconfidence=float(e3.get())
    print("minsupport={}".format(minsupport))
    print("minconfidence={}".format(minconfidence))
    choice=int(r.get())
    if choice==1:
        print("选择了Apriori算法")
        DoApriori()
        
    if choice==2:
        print("选择了fp-growth算法")
        DoFpGrowth()
    
    
    
def DoApriori():
    '''
    datalen=len(data)
    thisminsupport=minsupport/datalen
    '''
    itemsets, rules = apriori(data, minsupport,  minconfidence)
    print("频繁项集为{}".format(itemsets))
    print("关联规则为{}".format(rules))
    for item in itemsets:
        print(item)
    for rule in rules:
        print(rule)
    text_result.insert(END,"使用Apriori算法生成频繁项集和关联规则\n")
    text_result.insert(END,("频繁项集为：\n"+str(itemsets)+"\n"))
    text_result.insert(END,("关联规则为：\n"+str(rules)+"\n"))
   

def DoFpGrowth():
    strSet = set(functools.reduce(lambda a,b:a+b, data))
    strEncode = dict(zip(strSet,range(len(strSet)))) #编码字典，即:{'ArticleTag_BS': 6,'Country_Argentina': 53,etc...}
    strDecode = dict(zip(strEncode.values(), strEncode.keys()))  #解码字典，即:{6:'ArticleTag_BS',53:'Country_Argentina',etc...}
    data_int = [list(map(lambda item:strEncode[item],row)) for row in data]   #编码后的输入数据       
    itemsets=frequent_itemsets(data_int,.4)
    itemsets=dict(itemsets)
    #print(itemsets)
    rules = association_rules(itemsets, .5)   #这里设置置信度
    rules = list(rules)
    print("关联规则\n")
    print(rules)
    start=0
    tmpitem=[]
    for item in itemsets:
        tmpj=[]
        for j in item:
          j=strDecode[j]
          tmpj.append(j)
        tmpitem.append(set(tmpj))
    tmprules=[]
    for rule in rules:
        tmpj=[]
        for j in rule[:-2]:
           j=strDecode[list(j)[0]]
           tmpj.append(j)
        tmprules.append(tmpj)
    text_result.insert(END,"使用fp-growth算法生成频繁项集和关联规则\n")
    text_result.insert(END,("频繁项集为：\n"+str(tmpitem)+"\n"))
    text_result.insert(END,("关联规则为：\n"+str(tmprules)+"\n"))
    
def Clear():
    text_result.delete(1.0,'end')
    e1.delete(2.0,END)
    e2.delete(0, END)
    e3.delete(0, END)
    
def Close():
    root.quit()
  
def showtree():
    root.deiconify()
    
def save_file():
  file_path = filedialog.asksaveasfilename(title=u'保存文件')
  print('保存文件：', file_path)
  file_text = text_result.get('1.0',END)
  if file_path is not None:
    with open(file=file_path, mode='a+', encoding='utf-8') as file:
      file.write(file_text)
    dialog.Dialog(None, {'title': 'File Modified', 'text': '保存完成', 'bitmap': 'warning', 'default': 0,
               'strings': ('OK', 'Cancle')})
    print('保存完成')

#创建界面，进行事件绑定
root1 = Tk()
root1.geometry("850x450")
root1.title("关联规则挖掘系统")
root1.resizable=(False,False)
root1["background"] = "SkyBlue"
root1.attributes("-alpha", 0.9)

 
# Thinker总共提供了三种布局组件的方法：pack(),grid()和place()
# grid()方法允许你用表格的形式来管理组件的位置
# row选项代表行，column选项代表列
# 例如row=1，column=2表示第二行第三列(0表示第一行)

fram1=Frame(root1)
Label(fram1,text="频繁项集和关联规则",font=("华文行楷", 20), fg="green").grid(row=0,column=0)
text_result=Text(fram1)
text_result.grid(row=2,column=0)
fram1.grid(row=0,rowspan=4,column=4)

fram2=Frame(root1,width=300,height=500)

btnloaddata=Button(fram2, text="导入数据",command=LoadData)
btnloaddata.grid(row=0,column=0,pady=18)
Label(fram2, text="最小支持度").grid(row=1,column=0,pady=18)
Label(fram2, text="最小置信度").grid(row=2,column=0,pady=18)

e1_value="导入文件的路径:\n"
e1 = Text(fram2,width=20,height=3)
e1.insert(1.0,e1_value)


e2 = Entry(fram2)
e3 = Entry(fram2)


e1.grid(row=0, column=1,pady=18,padx=10)
e2.grid(row=1, column=1,pady=18,padx=10)
e3.grid(row=2, column=1,pady=18,padx=10)

fram2.grid(row=0,column=0,rowspan=4,columnspan=3,pady=18)

# 创建单选项
r = IntVar()
male_select = Radiobutton(root1,text="Apriori",value=1,variable=r)
male_select.grid(row=3,column=0)
female_select = Radiobutton(root1,text="Fp-growth",value=2,variable=r)
female_select.grid(row=3,column=1)

fram3=Frame(root1,width=100,height=10)

btnstart=Button(fram3,text="开始",command=Start)
btnclear=Button(fram3,text="清空",command=Clear)
btnshowtree=Button(fram3,text="显示Fptree",command=showtree)
btnsave=Button(fram3,text="保存结果",command=save_file)
#btnclose=Button(fram3,text="结束",command="")
btnstart.grid(row=0,column=0,padx=10,pady=10)
btnclear.grid(row=0,column=1,padx=10,pady=10)
btnshowtree.grid(row=0,column=2,padx=10,pady=10)
btnsave.grid(row=0,column=3,padx=10,pady=10)
#btnclose.grid(row=0,column=2,padx=10,pady=10)

fram3.grid(row=4,column=0,columnspan=3)

fram1["background"] = "white"
fram2["background"] = "SkyBlue"
fram3["background"] = "SkyBlue"


######################################################################################
# Example of embedding CEF Python browser using Tkinter toolkit.
# This example has two widgets: a navigation bar and a browser.
#
# NOTE: This example often crashes on Mac (Python 2.7, Tk 8.5/8.6)
#       during initial app loading with such message:
#       "Segmentation fault: 11". Reported as Issue #309.
#
# Tested configurations:
# - Tk 8.5 on Windows/Mac
# - Tk 8.6 on Linux
# - CEF Python v55.3+
#
# Known issue on Linux: When typing url, mouse must be over url
# entry widget otherwise keyboard focus is lost (Issue #255
# and Issue #284).

#以下代码为了嵌入html页面，具体代码没有仔细研究过，只是为了应用，只要修改embed_browser的url路径即可
from cefpython3 import cefpython as cef
import ctypes
try:
    import tkinter as tk
except ImportError:
    import Tkinter as tk
import sys
import os
import platform
import logging as _logging

global root
root = tk.Tk()

# Fix for PyCharm hints warnings
WindowUtils = cef.WindowUtils()

# Platforms
WINDOWS = (platform.system() == "Windows")
LINUX = (platform.system() == "Linux")
MAC = (platform.system() == "Darwin")

# Globals
logger = _logging.getLogger("tkinter_.py")

# Constants
# Tk 8.5 doesn't support png images
IMAGE_EXT = ".png" if tk.TkVersion > 8.5 else ".gif"


class MainFrame(tk.Frame):

    def __init__(self, root):
        self.browser_frame = None
        self.navigation_bar = None

        # Root
        root.geometry("900x640+300+200")
        tk.Grid.rowconfigure(root, 0, weight=1)
        tk.Grid.columnconfigure(root, 0, weight=1)

        # MainFrame
        tk.Frame.__init__(self, root)
        self.master.title("FpTree")
        self.master.protocol("WM_DELETE_WINDOW", self.on_close)
        self.master.bind("<Configure>", self.on_root_configure)
        self.setup_icon()
        self.bind("<Configure>", self.on_configure)
        self.bind("<FocusIn>", self.on_focus_in)
        self.bind("<FocusOut>", self.on_focus_out)

        # NavigationBar
        self.navigation_bar = NavigationBar(self)
        self.navigation_bar.grid(row=0, column=0,
                                 sticky=(tk.N + tk.S + tk.E + tk.W))
        tk.Grid.rowconfigure(self, 0, weight=0)
        tk.Grid.columnconfigure(self, 0, weight=0)

        # BrowserFrame
        self.browser_frame = BrowserFrame(self, self.navigation_bar)
        self.browser_frame.grid(row=1, column=0,
                                sticky=(tk.N + tk.S + tk.E + tk.W))
        tk.Grid.rowconfigure(self, 1, weight=1)
        tk.Grid.columnconfigure(self, 0, weight=1)

        # Pack MainFrame
        self.pack(fill=tk.BOTH, expand=tk.YES)

    def on_root_configure(self, _):
        logger.debug("MainFrame.on_root_configure")
        if self.browser_frame:
            self.browser_frame.on_root_configure()

    def on_configure(self, event):
        logger.debug("MainFrame.on_configure")
        if self.browser_frame:
            width = event.width
            height = event.height
            if self.navigation_bar:
                height = height - self.navigation_bar.winfo_height()
            self.browser_frame.on_mainframe_configure(width, height)

    def on_focus_in(self, _):
        logger.debug("MainFrame.on_focus_in")

    def on_focus_out(self, _):
        logger.debug("MainFrame.on_focus_out")

    def on_close(self):
        if self.browser_frame:
            self.browser_frame.on_root_close()
        self.master.destroy()

    def get_browser(self):
        if self.browser_frame:
            return self.browser_frame.browser
        return None

    def get_browser_frame(self):
        if self.browser_frame:
            return self.browser_frame
        return None

    def setup_icon(self):
        resources = os.path.join(os.path.dirname(__file__), "resources")
        icon_path = os.path.join(resources, "tkinter"+IMAGE_EXT)
        if os.path.exists(icon_path):
            self.icon = tk.PhotoImage(file=icon_path)
            # noinspection PyProtectedMember
            self.master.call("wm", "iconphoto", self.master._w, self.icon)


class BrowserFrame(tk.Frame):

    def __init__(self, master, navigation_bar=None):
        self.navigation_bar = navigation_bar
        self.closing = False
        self.browser = None
        tk.Frame.__init__(self, master)
        self.bind("<FocusIn>", self.on_focus_in)
        self.bind("<FocusOut>", self.on_focus_out)
        self.bind("<Configure>", self.on_configure)
        self.focus_set()

    def embed_browser(self):
        window_info = cef.WindowInfo()
        rect = [0, 0, self.winfo_width(), self.winfo_height()]
        window_info.SetAsChild(self.get_window_handle(), rect)
        self.browser = cef.CreateBrowserSync(window_info,
                                             url="file:///fptree.html") #todo  
        #只需要修改url中的内容，嵌入本地html
        assert self.browser
        self.browser.SetClientHandler(LoadHandler(self))
        self.browser.SetClientHandler(FocusHandler(self))
        self.message_loop_work()

    def get_window_handle(self):
        if self.winfo_id() > 0:
            return self.winfo_id()
        elif MAC:
            # On Mac window id is an invalid negative value (Issue #308).
            # This is kind of a dirty hack to get window handle using
            # PyObjC package. If you change structure of windows then you
            # need to do modifications here as well.
            # noinspection PyUnresolvedReferences
            from AppKit import NSApp
            # noinspection PyUnresolvedReferences
            import objc
            # Sometimes there is more than one window, when application
            # didn't close cleanly last time Python displays an NSAlert
            # window asking whether to Reopen that window.
            # noinspection PyUnresolvedReferences
            return objc.pyobjc_id(NSApp.windows()[-1].contentView())
        else:
            raise Exception("Couldn't obtain window handle")

    def message_loop_work(self):
        cef.MessageLoopWork()
        self.after(10, self.message_loop_work)

    def on_configure(self, _):
        if not self.browser:
            self.embed_browser()

    def on_root_configure(self):
        # Root <Configure> event will be called when top window is moved
        if self.browser:
            self.browser.NotifyMoveOrResizeStarted()

    def on_mainframe_configure(self, width, height):
        if self.browser:
            if WINDOWS:
                ctypes.windll.user32.SetWindowPos(
                    self.browser.GetWindowHandle(), 0,
                    0, 0, width, height, 0x0002)
            elif LINUX:
                self.browser.SetBounds(0, 0, width, height)
            self.browser.NotifyMoveOrResizeStarted()

    def on_focus_in(self, _):
        logger.debug("BrowserFrame.on_focus_in")
        if self.browser:
            self.browser.SetFocus(True)

    def on_focus_out(self, _):
        logger.debug("BrowserFrame.on_focus_out")
        if self.browser:
            self.browser.SetFocus(False)

    def on_root_close(self):
        if self.browser:
            self.browser.CloseBrowser(True)
            self.clear_browser_references()
        self.destroy()

    def clear_browser_references(self):
        # Clear browser references that you keep anywhere in your
        # code. All references must be cleared for CEF to shutdown cleanly.
        self.browser = None


class LoadHandler(object):

    def __init__(self, browser_frame):
        self.browser_frame = browser_frame

    def OnLoadStart(self, browser, **_):
        if self.browser_frame.master.navigation_bar:
            self.browser_frame.master.navigation_bar.set_url(browser.GetUrl())


class FocusHandler(object):

    def __init__(self, browser_frame):
        self.browser_frame = browser_frame

    def OnTakeFocus(self, next_component, **_):
        logger.debug("FocusHandler.OnTakeFocus, next={next}"
                     .format(next=next_component))

    def OnSetFocus(self, source, **_):
        logger.debug("FocusHandler.OnSetFocus, source={source}"
                     .format(source=source))
        return False

    def OnGotFocus(self, **_):
        """Fix CEF focus issues (#255). Call browser frame's focus_set
           to get rid of type cursor in url entry widget."""
        logger.debug("FocusHandler.OnGotFocus")
        self.browser_frame.focus_set()


class NavigationBar(tk.Frame):
    def __init__(self, master):
        self.back_state = tk.NONE
        self.forward_state = tk.NONE
        self.back_image = None
        self.forward_image = None
        self.reload_image = None

        tk.Frame.__init__(self, master)
        resources = os.path.join(os.path.dirname(__file__), "resources")

        # Back button
        back_png = os.path.join(resources, "back"+IMAGE_EXT)
        if os.path.exists(back_png):
            self.back_image = tk.PhotoImage(file=back_png)
        self.back_button = tk.Button(self, image=self.back_image,
                                     command=self.go_back)
        self.back_button.grid(row=0, column=0)

        # Forward button
        forward_png = os.path.join(resources, "forward"+IMAGE_EXT)
        if os.path.exists(forward_png):
            self.forward_image = tk.PhotoImage(file=forward_png)
        self.forward_button = tk.Button(self, image=self.forward_image,
                                        command=self.go_forward)
        self.forward_button.grid(row=0, column=1)

        # Reload button
        reload_png = os.path.join(resources, "reload"+IMAGE_EXT)
        if os.path.exists(reload_png):
            self.reload_image = tk.PhotoImage(file=reload_png)
        self.reload_button = tk.Button(self, image=self.reload_image,
                                       command=self.reload)
        self.reload_button.grid(row=0, column=2)

        # Url entry
        self.url_entry = tk.Entry(self)
        self.url_entry.bind("<FocusIn>", self.on_url_focus_in)
        self.url_entry.bind("<FocusOut>", self.on_url_focus_out)
        self.url_entry.bind("<Return>", self.on_load_url)
        self.url_entry.bind("<Button-1>", self.on_button1)
        self.url_entry.grid(row=0, column=3,
                            sticky=(tk.N + tk.S + tk.E + tk.W))
        tk.Grid.rowconfigure(self, 0, weight=100)
        tk.Grid.columnconfigure(self, 3, weight=100)

        # Update state of buttons
        self.update_state()

    def go_back(self):
        if self.master.get_browser():
            self.master.get_browser().GoBack()

    def go_forward(self):
        if self.master.get_browser():
            self.master.get_browser().GoForward()

    def reload(self):
        if self.master.get_browser():
            self.master.get_browser().Reload()

    def set_url(self, url):
        self.url_entry.delete(0, tk.END)
        self.url_entry.insert(0, url)

    def on_url_focus_in(self, _):
        logger.debug("NavigationBar.on_url_focus_in")

    def on_url_focus_out(self, _):
        logger.debug("NavigationBar.on_url_focus_out")

    def on_load_url(self, _):
        if self.master.get_browser():
            self.master.get_browser().StopLoad()
            self.master.get_browser().LoadUrl(self.url_entry.get())

    def on_button1(self, _):
        """Fix CEF focus issues (#255). See also FocusHandler.OnGotFocus."""
        logger.debug("NavigationBar.on_button1")
        self.master.master.focus_force()

    def update_state(self):
        browser = self.master.get_browser()
        if not browser:
            if self.back_state != tk.DISABLED:
                self.back_button.config(state=tk.DISABLED)
                self.back_state = tk.DISABLED
            if self.forward_state != tk.DISABLED:
                self.forward_button.config(state=tk.DISABLED)
                self.forward_state = tk.DISABLED
            self.after(100, self.update_state)
            return
        if browser.CanGoBack():
            if self.back_state != tk.NORMAL:
                self.back_button.config(state=tk.NORMAL)
                self.back_state = tk.NORMAL
        else:
            if self.back_state != tk.DISABLED:
                self.back_button.config(state=tk.DISABLED)
                self.back_state = tk.DISABLED
        if browser.CanGoForward():
            if self.forward_state != tk.NORMAL:
                self.forward_button.config(state=tk.NORMAL)
                self.forward_state = tk.NORMAL
        else:
            if self.forward_state != tk.DISABLED:
                self.forward_button.config(state=tk.DISABLED)
                self.forward_state = tk.DISABLED
        self.after(100, self.update_state)

def Click1():
    root.deiconify()

if __name__ == '__main__':
    logger.setLevel(_logging.INFO)
    stream_handler = _logging.StreamHandler()
    formatter = _logging.Formatter("[%(filename)s] %(message)s")
    stream_handler.setFormatter(formatter)
    logger.addHandler(stream_handler)
    logger.info("CEF Python {ver}".format(ver=cef.__version__))
    logger.info("Python {ver} {arch}".format(
            ver=platform.python_version(), arch=platform.architecture()[0]))
    logger.info("Tk {ver}".format(ver=tk.Tcl().eval('info patchlevel')))
    assert cef.__version__ >= "55.3", "CEF Python v55.3+ required to run this"
    sys.excepthook = cef.ExceptHook  # To shutdown all CEF processes on error
    root.withdraw()
    #root = tk.Tk()
   
    app = MainFrame(root)
    # Tk must be initialized before CEF otherwise fatal error (Issue #306)
    cef.Initialize()
    root1.mainloop()
    app.mainloop()
    cef.Shutdown()