yt_stats.py

import tkinter as tk
import customtkinter as ctk
import matplotlib.pyplot as plt
from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg # enbales to place the  graph into my window
plt.style.use("dark_background")

# CHANNELS
# [0] category = 'Film and Animation'
# [1] join_date = '2017-05-21'
# [2] channel_id = 'UCBJuEqXfXTdcPSbGO9qqn1g'
# [3] channel_name = 'MagnusNation'
# [4] subscribers = '65100'
# [5] videos = '28'
# [6] subscriber_rank = '231223.0'

# TIMESERIES
# [0] channel_id
# [1] category
# [2] datetime
# [3] views
# [4] delta_views
# [5] subscribers
# [6] delta_subscribers
# [7] videos
# [8] delta_videos
# [9] activity

class Data_analyzer:
    def __init__(self):
        self.main_window = ctk.CTk()
        self.main_window.geometry("1280x720")

        self.left_frame = ctk.CTkFrame(self.main_window, width=230, fg_color="#100f0f", corner_radius=0)
        self.left_frame.pack(side="left", fill="y")

        self.middle_frame = ctk.CTkFrame(self.main_window, width=1, fg_color="#575653", border_width=1, border_color="#575653", corner_radius=0)
        self.middle_frame.pack(side="left", fill="y")

        self.right_frame = ctk.CTkFrame(self.main_window, fg_color="#1C1B1A", corner_radius=0)
        self.right_frame.pack(side="left", fill="both", expand=True)

    def clear_window(self):
        keep = [self.overview_button, self.leaderboard_button, self.categories_button, self.subgrowth_button, self.left_frame, self.middle_frame, self.right_frame,
                self.general_label, self.analysis_label, self.efficiency_button, self.outliers_button, self.tools_label, self.rank_button, self.scatter_button]
        for w in self.main_window.winfo_children(): # goes through 'widgets' created insie main_window
            if w not in keep:
                w.destroy()

        plt.close('all')

    def style_ax(self, ax, fig):
        ax.tick_params(axis='x', colors='#FFFCF0')
        ax.tick_params(axis='y', colors='#FFFCF0')
        ax.spines['bottom'].set_color('#CECDC3')
        ax.spines['left'].set_color('#CECDC3')
        ax.spines['top'].set_color('#CECDC3')
        ax.spines['right'].set_color('#CECDC3')
        fig.patch.set_facecolor('#100f0f')
        ax.set_facecolor('#100f0f')

    def main_menu(self):
        # TEXT
        self.general_label = ctk.CTkLabel(self.main_window, text="General", fg_color="#100f0f", font = ('', 15), text_color="#878580")
        self.general_label.place(x=15, y=10)

        # MENU BUTTONS
        self.overview_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Overview', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.overview_menu()) 
        self.overview_button.place(x=15, y=45)

        self.leaderboard_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Leaderboard', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.leaderboard_menu()) 
        self.leaderboard_button.place(x=15, y=90)

        self.categories_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Categories', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.categories_menu()) 
        self.categories_button.place(x=15, y=135)

        # TEXT
        self.analysis_label = ctk.CTkLabel(self.main_window, text="Analysis", fg_color="#100f0f", font = ('', 15), text_color="#878580")
        self.analysis_label.place(x=15, y=195)

        self.subgrowth_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Subscriber Growth', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.subgrowth_menu()) 
        self.subgrowth_button.place(x=15, y=230)

        self.efficiency_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Efficiency', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.efficiency_menu()) 
        self.efficiency_button.place(x=15, y=275)

        self.outliers_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Outliers', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.outliers_menu()) 
        self.outliers_button.place(x=15, y=320)

        self.scatter_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Age vs Success', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.agevssuccess_menu()) 
        self.scatter_button.place(x=15, y=365)

        # TEXT
        self.tools_label = ctk.CTkLabel(self.main_window, text="Tools", fg_color="#100f0f", font = ('', 15), text_color="#878580")
        self.tools_label.place(x=15, y=425)

        self.rank_button = ctk.CTkButton(self.main_window, height=40, width=200, text = 'Rank Yourself', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", anchor='w', font = ('', 15),
                                         command = lambda: self.rank_menu()) 
        self.rank_button.place(x=15, y=460)

    def overview_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Overview", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label = ctk.CTkLabel(self.main_window, text="Data available - 2016 to 2019", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label.place(x=260, y=65)

        # INFO
        ## CHANNEL AMOUNT
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=260, y=120)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        channel_amount_label = ctk.CTkLabel(frame, text="Total Channels", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        channel_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

        channel_amount = 0
        for row in self.channels:
            channel_amount += 1

        channel_amount_label = ctk.CTkLabel(frame, text=str(channel_amount), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
        channel_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

        ## CATEGORY AMOUNT
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=525, y=120)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        categories_amount_label = ctk.CTkLabel(frame, text="Categories", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        categories_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

        categories_count_list = []
        categories_count = 0
        for row in self.channels:
            parts = row.split('\t')

            if parts[0] not in categories_count_list:
                categories_count_list.append(parts[0])
                categories_count += 1

        categories_amount_label = ctk.CTkLabel(frame, text=str(categories_count), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
        categories_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

        ## AVG SUBS
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=790, y=120)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        subs_amount_label = ctk.CTkLabel(frame, text="Avg. Subscribers", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        subs_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

        row_count = 0
        sub_storage = 0
        for row in self.channels:
            parts = row.split('\t')
            row_count += 1
            sub_storage += int(parts[4])

        subs_amount_label = ctk.CTkLabel(frame, text=str(self.letterform(sub_storage//row_count)), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
        subs_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

        s.overview_bargraph()

    def leaderboard_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Leaderbord", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label = ctk.CTkLabel(self.main_window, text="Shows the top channels based on subscriber count during this period", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label.place(x=260, y=65)

        frame = ctk.CTkFrame(self.main_window, width=900, height=50, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=260, y=120)

        frame = ctk.CTkFrame(self.main_window, width=900, height=450, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=260, y=180)

        y_gain = 133
        for i in range (10 + 1):
            channels = self.channels[i-1]
            parts = channels.split('\t')

            if i == 0:
                name = 'Channel'
                category = 'Category'
                subscribers = 'Subs'
                videos = 'Videos'
                join_date = 'Join Date'
            else:
                name = parts[3]
                category = parts[0]
                subscribers = self.letterform(parts[4])
                videos = self.letterform(parts[5])
                join_date = parts[1][:4]

                channel_amount_label = ctk.CTkLabel(self.main_window, text=str(i), fg_color="#100f0f", font = ('Courier New', 20), text_color="#FFFCF0")
                channel_amount_label.place(x=280, y = y_gain)

            channel_amount_label = ctk.CTkLabel(self.main_window, text=name, fg_color="#100f0f", font = ('Courier New', 20, 'bold'), text_color="#FFFCF0")
            channel_amount_label.place(x=320, y = y_gain)

            channel_amount_label = ctk.CTkLabel(self.main_window, text=category, fg_color="#100f0f", font = ('Courier New', 20), text_color="#FFFCF0")
            channel_amount_label.place(x=620, y = y_gain)

            channel_amount_label = ctk.CTkLabel(self.main_window, text=subscribers, fg_color="#100f0f", font = ('Courier New', 20), text_color="#FFFCF0")
            channel_amount_label.place(x=820, y = y_gain)

            channel_amount_label = ctk.CTkLabel(self.main_window, text=videos, fg_color="#100f0f", font = ('Courier New', 20), text_color="#FFFCF0")
            channel_amount_label.place(x=920, y = y_gain)

            channel_amount_label = ctk.CTkLabel(self.main_window, text=join_date, fg_color="#100f0f", font = ('Courier New', 20), text_color="#FFFCF0")
            channel_amount_label.place(x=1020, y = y_gain)

            if i == 0:
                y_gain += 63
            else:
                y_gain += 30

    def categories_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Category Representation", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label = ctk.CTkLabel(self.main_window, text="Pie graph shows how each category is represented on YouTube", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label.place(x=260, y=65)

        s.categories_piegraph('Category Representation')

        # INFO
        ## MOST COMMON
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=990, y=120)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        common_amount_label = ctk.CTkLabel(frame, text="Most Common", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        common_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

        common_amount_label = ctk.CTkLabel(frame, text=self.most_common_category_name, fg_color="#100f0f", font = ('Courier New', 25, 'bold'), text_color="#FFFCF0")
        common_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

        ## AVG
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=990, y=260)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        row_count = 0
        sub_storage = 0
        video_storage = 0
        for row in self.channels:
            parts = row.split('\t')
            if parts[0] == self.most_common_category_name:
                row_count += 1
                sub_storage += int(parts[4])
                video_storage += int(parts[5])


        subavg_label = ctk.CTkLabel(frame, text="Total Subscribers", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        subavg_label.pack(anchor='w', padx=20, pady=(15, 0))

        subavg_label = ctk.CTkLabel(frame, text=str(self.letterform(sub_storage)), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
        subavg_label.pack(anchor='w', padx=20, pady=(0, 20))

        ## Channel amount
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=990, y=400)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        channel_amount_label = ctk.CTkLabel(frame, text="Channels", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        channel_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

        channel_amount_label = ctk.CTkLabel(frame, text=str(row_count), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
        channel_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

        ## Video amount
        frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=990, y=540)
        frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

        channel_amount_label = ctk.CTkLabel(frame, text="Total Videos", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
        channel_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

        channel_amount_label = ctk.CTkLabel(frame, text=str(self.letterform(video_storage)), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
        channel_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

    def subgrowth_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Subscriber Growth", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label = ctk.CTkLabel(self.main_window, text="Line graph shows subscriber growth from 2016 to 2019", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label.place(x=260, y=65)

        listbox_frame = ctk.CTkFrame(self.main_window, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        listbox_frame.place(x=260, y=120)

        self.load_channel_listbox(listbox_frame)

        self.channel_listbox.bind("<<ListboxSelect>>", lambda event: self.subgrowth_linegraph(5, 2, 'Subscribers vs Date')) # graph creation upon clicking channel in the listbox wihtout any external input

        self.channel_listbox.selection_set(0, 1)
        self.subgrowth_linegraph(5, 2, 'Subscribers vs Date') # empty graph so there isn't blank space

    def efficiency_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Efficiency", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label = ctk.CTkLabel(self.main_window, text="How efficient are channels based on the amount of views per video", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label.place(x=260, y=65)

        listbox_frame = ctk.CTkFrame(self.main_window, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        listbox_frame.place(x=260, y=120)

        self.load_channel_listbox(listbox_frame)

        self.channel_listbox.bind("<<ListboxSelect>>", lambda event: self.channel_efficiency_linegraph(7, 3, 'Views vs Videos')) # graph creation upon clicking channel in the listbox wihtout any external input

        self.channel_listbox.selection_set(0, 2)
        self.channel_efficiency_linegraph(7, 3, 'Views vs Videos') # eliminated blank space

    def outliers_menu(self):
        self.clear_window()

        # TITLES:
        title_label = ctk.CTkLabel(self.main_window, text="Outliers", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label_2 = ctk.CTkLabel(self.main_window, text="Over-performing channels, ranking equation lies in subscriber amount divided by total video amount",
                                     fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label_2.place(x=260, y=65)

        listbox_frame = ctk.CTkFrame(self.main_window, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        listbox_frame.place(x=260, y=120)

        self.load_categories_listbox(listbox_frame)

        def outliers_output():
            keep = [self.overview_button, self.leaderboard_button, self.categories_button, self.subgrowth_button, self.left_frame, self.middle_frame, self.right_frame,
                    self.general_label, self.analysis_label, self.efficiency_button, self.outliers_button, self.tools_label, self.rank_button,
                    listbox_frame, title_label, title_label_2, self.scatter_button]
            for w in self.main_window.winfo_children(): # goes through 'widgets' created insie main_window
                if w not in keep:
                    w.destroy()

            efficiency_value = 0
            for row in self.channels:
                parts = row.split('\t')

                if parts[0] == self.category_listbox.get(self.category_listbox.curselection()):
                    efficiency = int(parts[4])//int(parts[5])
                    if efficiency > efficiency_value:
                        efficiency_value = efficiency
                        top_performer = parts[3]

            # CATEGORY NAME
            frame = ctk.CTkFrame(self.main_window, width=700, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
            frame.place(x=510, y=120)
            frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui
            
            category_label = ctk.CTkLabel(frame, text="Category", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
            category_label.pack(anchor='w', padx=20, pady=(15, 0))

            cateogry_label = ctk.CTkLabel(frame, text=self.category_listbox.get(self.category_listbox.curselection()),
                                                fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
            cateogry_label.pack(anchor='w', padx=20, pady=(0, 20))

            # OUTLIER NAME
            frame = ctk.CTkFrame(self.main_window, width = 435, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
            frame.place(x=510, y=255)
            frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

            outlier_label = ctk.CTkLabel(frame, text="Top Performer", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
            outlier_label.pack(anchor='w', padx=20, pady=(15, 0))

            outlier_label = ctk.CTkLabel(frame, text=top_performer,
                                                fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
            outlier_label.pack(anchor='w', padx=20, pady=(0, 20))

            for row in self.channels:
                parts = row.split('\t')

                if parts[3] == top_performer:
                    subscribers = parts[4]
                    videos = parts[5]
                    join_date = parts[1][:12]

            # EXTRAS
            frame = ctk.CTkFrame(self.main_window, width=435, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
            frame.place(x=775, y=390)
            frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

            common_amount_label = ctk.CTkLabel(frame, text='Join Date', fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
            common_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

            common_amount_label = ctk.CTkLabel(frame, text=join_date, fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
            common_amount_label.pack(anchor='w', padx=20, pady=(0, 20))


            frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
            frame.place(x=960, y=255)
            frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

            subscriber_amount_label = ctk.CTkLabel(frame, text='Subscribers', fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
            subscriber_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

            subscriber_amount_label = ctk.CTkLabel(frame, text=self.letterform(subscribers), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
            subscriber_amount_label.pack(anchor='w', padx=20, pady=(0, 20))


            frame = ctk.CTkFrame(self.main_window, width=250, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
            frame.place(x=510, y=390)
            frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

            video_amount_label = ctk.CTkLabel(frame, text='Videos', fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
            video_amount_label.pack(anchor='w', padx=20, pady=(15, 0))

            video_amount_label = ctk.CTkLabel(frame, text=self.letterform(videos), fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
            video_amount_label.pack(anchor='w', padx=20, pady=(0, 20))

            plt.close('all') # now the graphs wont be on top of each other and multiple, but just one
            fig, ax = plt.subplots()
            fig.subplots_adjust(left=-0.045, right=1.045)
            x_value = []
            y_value = []

            for row in self.timeseries:
                parts = row.split('\t')
                
                # Check just the ones selected in the ListBox
                if top_performer == self.channel_dict[parts[0]]:
                    y_value.append(float(parts[3]))
                    x_value.append(parts[3][:10]) # also remove hour and minute, [start:stop]

            ax.fill_between(x_value, y_value, color = '#1C1B1A', edgecolor = '#878580')
            ax.set_xticks(x_value[::35]) # [start:stop:step]

            frame = ctk.CTkFrame(self.main_window, width=700, height=165, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
            frame.place(x=510, y=525)
            frame.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

            ax.tick_params(axis='x', colors='#100f0f')
            ax.tick_params(axis='y', colors='#100f0f')
            ax.spines['bottom'].set_color('#100f0f')
            ax.spines['left'].set_color('#100f0f')
            ax.spines['top'].set_color('#100f0f')
            ax.spines['right'].set_color('#100f0f')
            fig.patch.set_facecolor('#100f0f')
            ax.set_facecolor('#100f0f')

            # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
            canvas = FigureCanvasTkAgg(fig, master=frame)
            widget = canvas.get_tk_widget()
            # fig.subplots_adjust(left = 0.05, right = 0.95, bottom = 0.05, top = 0.95) # from 0 to 1, from start to end of the window
            widget.pack(padx=5, pady=5) # pad so rounded corners work

        self.category_listbox.bind("<<ListboxSelect>>", lambda event: outliers_output()) # graph creation upon clicking channel in the listbox wihtout any external input

        # pre-select first item
        self.category_listbox.selection_set(13)
        outliers_output()

    def agevssuccess_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Does Age Matter?", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label = ctk.CTkLabel(self.main_window, text="Mind the fact that many channels didn't upload for years after the initial creation", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label.place(x=260, y=65)

        listbox_frame = ctk.CTkFrame(self.main_window, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        listbox_frame.place(x=260, y=120)

        self.load_channel_listbox(listbox_frame)

        self.channel_listbox.bind("<<ListboxSelect>>", lambda event: self.agevssuccess_scattergraph(4, 1, 'Subscribers vs Join Date')) # graph creation upon clicking channel in the listbox wihtout any external input

        self.channel_listbox.selection_set(0, 5)
        self.agevssuccess_scattergraph(4, 1, 'Subscribers vs Join Date') # empty graph so there isn't blank space

    def rank_menu(self):
        self.clear_window()
        # TITLES
        title_label = ctk.CTkLabel(self.main_window, text="Rank Yourself", fg_color="#1C1B1A", font = ('Courier New', 40, 'bold'), text_color="#FFFCF0")
        title_label.place(x=260, y=25)

        title_label_2 = ctk.CTkLabel(self.main_window, text="Enter your channel statistics to see where you would rank in the dataset", fg_color="#1C1B1A", font = ('Courier New', 15), text_color="#878580")
        title_label_2.place(x=260, y=65)

        frame = ctk.CTkFrame(self.main_window, width=700, height=100, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame.place(x=260, y=120)
        
        categories_list = []
        for row in self.channels:
            parts = row.split('\t')
            if parts[0] not in categories_list:
                categories_list.append(parts[0])

        title_label_3 = ctk.CTkLabel(self.main_window, text="Category", fg_color="#100f0f", font = ('Courier New', 16), text_color="#878580")
        title_label_3.place(x=295, y = 140)

        category_combo = ctk.CTkComboBox(frame, values=categories_list, width = 200, height = 40, state = 'readonly', font = ('', 18, 'bold'),
                                fg_color="#1C1B1A",
                                border_color="#575653",
                                dropdown_fg_color="#282726",
                                button_color="#575653") 
        category_combo.pack(side = 'left', padx = (30, 30), pady = (50, 30))

        title_label_4 = ctk.CTkLabel(self.main_window, text="Subscribers", fg_color="#100f0f", font = ('Courier New', 16), text_color="#878580")
        title_label_4.place(x=525, y = 140)

        subs_entry = ctk.CTkEntry(frame, width=200, height=40, placeholder_text='eg. 9000000', font = ('', 18, 'bold'),
                                  fg_color="#1C1B1A",
                                  border_color="#575653")
        subs_entry.pack(side = 'left', padx = (0, 30), pady = (50, 30))

        title_label_5 = ctk.CTkLabel(self.main_window, text="Videos", fg_color="#100f0f", font = ('Courier New', 16), text_color="#878580")
        title_label_5.place(x=755, y = 140)

        video_entry = ctk.CTkEntry(frame, width=200, height=40, placeholder_text='eg. 110', font = ('', 18, 'bold'),
                                   fg_color="#1C1B1A",
                                   border_color="#575653")
        video_entry.pack(side = 'left', padx = (0, 30), pady = (50, 30))

        # ENTRY BUTTON
        frame2 = ctk.CTkFrame(self.main_window, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame2.place(x=1010, y=120)

        rank_button = ctk.CTkButton(frame2, height=60, width=150, text = 'RANK', corner_radius=8,
                                         border_width=1, border_color='#575653', fg_color='#1C1B1A', bg_color="#100f0f", hover_color="#282726", font = ('', 40, 'bold'),
                                         command = lambda: rank_output()) 
        rank_button.pack(padx = 30, pady = 30)

        def rank_output():
            # isdigit does not allow negatives or decimals which is nice
            if category_combo.get() != '' and subs_entry.get().isdigit() and video_entry.get().isdigit() and int(subs_entry.get()) > 0 and int(video_entry.get()) > 0:
                keep = [self.overview_button, self.leaderboard_button, self.categories_button, self.subgrowth_button, self.left_frame, self.middle_frame, self.right_frame,
                        self.general_label, self.analysis_label, self.efficiency_button, self.outliers_button, self.tools_label, self.rank_button,
                        self.scatter_button, category_combo, subs_entry, video_entry, frame, frame2, title_label, title_label_2, title_label_3, title_label_4, title_label_5, rank_button]
                for w in self.main_window.winfo_children(): # goes through 'widgets' created insie main_window
                    if w not in keep:
                        w.destroy()

                higher_subs_cat = 0
                all_subs_cat = 0
                higher_subs = 0
                all_subs = 0
                for row in self.channels:
                    parts = row.split('\t')

                    all_subs +=1
                    if int(parts[4]) > int(subs_entry.get()):
                        higher_subs +=1

                    if parts[0] == category_combo.get():
                        if int(parts[4]) > int(subs_entry.get()):
                            higher_subs_cat += 1

                        all_subs_cat += 1 

                frame3 = ctk.CTkFrame(self.main_window, width=300, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
                frame3.place(x=260, y=270)
                frame3.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

                subscriber_rank_label = ctk.CTkLabel(frame3, text=f"Subscriber Rank in {category_combo.get()}", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
                subscriber_rank_label.pack(anchor='w', padx=20, pady=(15, 0))

                subscriber_rank_label = ctk.CTkLabel(frame3, text=f'Top {int((higher_subs_cat/all_subs_cat) * 100)}%', fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
                subscriber_rank_label.pack(anchor='w', padx=20, pady=(0, 20))

                frame3 = ctk.CTkFrame(self.main_window, width=300, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
                frame3.place(x=590, y=270)
                frame3.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

                subscriber_rank_label = ctk.CTkLabel(frame3, text="Overall Subscriber Rank", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
                subscriber_rank_label.pack(anchor='w', padx=20, pady=(15, 0))

                subscriber_rank_label = ctk.CTkLabel(frame3, text=f'Top {int((higher_subs/all_subs) * 100)}%', fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
                subscriber_rank_label.pack(anchor='w', padx=20, pady=(0, 20))

                higher_eff = 0
                all_eff = 0
                for row in self.channels:
                    parts = row.split('\t')

                    if int(parts[5])/int(parts[4]) < int(video_entry.get())/int(subs_entry.get()):
                        higher_eff += 1

                    all_eff += 1

                frame3 = ctk.CTkFrame(self.main_window, width=300, height=120, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
                frame3.place(x=920, y=270)
                frame3.pack_propagate(False) # insides can be packed without changing the frame size to have consistency in the gui

                subscriber_rank_label = ctk.CTkLabel(frame3, text="Video Amount Efficiency", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
                subscriber_rank_label.pack(anchor='w', padx=20, pady=(15, 0))

                subscriber_rank_label = ctk.CTkLabel(frame3, text=f'Top {int((higher_eff/all_eff) * 100)}%', fg_color="#100f0f", font = ('Courier New', 50, 'bold'), text_color="#FFFCF0")
                subscriber_rank_label.pack(anchor='w', padx=20, pady=(0, 20))

                self.categories_linegraph(category_combo.get())

                subscriber_rank_label = ctk.CTkLabel(self.main_window, text="Subscriber Performance Over Time for Your Category", fg_color="#100f0f", font = ('Courier New', 15), text_color="#878580")
                subscriber_rank_label.place(x=280, y = 435)

    def categories_linegraph(self, category):
        fig, ax = plt.subplots()
        fig.subplots_adjust(left=0, right=1)

        frame3 = ctk.CTkFrame(self.main_window, width=960, height=270, fg_color="#100f0f", bg_color='#1C1B1A', corner_radius=12)
        frame3.place(x=260, y=420)
        frame3.pack_propagate(False)
        

        y_value = []
        x_value_dict = {}
        x_value = []
    
        for row in self.timeseries:
            parts = row.split('\t')

            if parts[1] == category:
                if parts[2][:10] not in x_value:
                    x_value_dict[parts[2][:10]] = 0

                x_value_dict[parts[2][:10]] += float(parts[5])

        x_value_dict = dict(sorted(x_value_dict.items())) # sorts the items by date becuase of the inconsistent amount od data

        for key in x_value_dict:
            x_value.append(key)
            y_value.append(int(x_value_dict[key]))

        ax.fill_between(x_value, y_value, color = '#1C1B1A', edgecolor = '#878580')

        ax.tick_params(axis='x', colors='#100f0f')
        ax.tick_params(axis='y', colors='#100f0f')
        ax.spines['bottom'].set_color('#100f0f')
        ax.spines['left'].set_color('#100f0f')
        ax.spines['top'].set_color('#100f0f')
        ax.spines['right'].set_color('#100f0f')
        fig.patch.set_facecolor('#100f0f')
        ax.set_facecolor('#100f0f')

        # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
        canvas = FigureCanvasTkAgg(fig, master=frame3)
        widget = canvas.get_tk_widget()
        widget.pack(fill = 'both', padx=5, pady=5) # pad so rounded corners work

    def overview_bargraph(self):
        frame = ctk.CTkFrame(
            self.main_window,
            fg_color="#100f0f",
            bg_color="#1C1B1A",
            corner_radius=12
        )

        frame.place(x=260, y=270)

        categories_subs_dict = {}
        categories_count_dict = {}

        for row in self.channels:
            parts = row.split('\t')

            if parts[0] not in categories_subs_dict:
                categories_subs_dict[parts[0]] = 0
                categories_count_dict[parts[0]] = 0

            categories_subs_dict[parts[0]] += int(parts[4])
            categories_count_dict[parts[0]] += 1

        category_name = []
        category_subs = []

        for category in categories_subs_dict:
            avg = categories_subs_dict[category] // categories_count_dict[category]

            if categories_count_dict[category] > 30:
                if len(category_name) < 12:
                    category_name.append(category)
                    category_subs.append(avg)

        fig, ax = plt.subplots(figsize=(9, 4.1))
        ax.barh(category_name, category_subs, color = '#1C1B1A', edgecolor = '#878580')
        fig.subplots_adjust(left = 0.18, right = 0.95, bottom = 0.14) # from 0 to 1, from start to end of the window

        ax.set_title("Categories vs Subscribers", color='#FFFCF0')

        self.style_ax(ax, fig)

        # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
        canvas = FigureCanvasTkAgg(fig, master=frame)
        widget = canvas.get_tk_widget()
        widget.pack(padx=5, pady=5) # pad so rounded corners work

    def categories_piegraph(self, title):
        plt.close('all') # now the graphs wont be on top of each other and multiple, but just one
        fig, ax = plt.subplots(figsize=(7, 5.32))

        categories_count_dict = {}
        categories_count_list = []
        categories_name_list = []

        for row in self.channels:
            parts = row.split('\t')

            if parts[0] not in categories_count_dict:
                categories_count_dict[parts[0]] = 0

        categories_count_dict['Others'] = 0 # Add Others to accomodate for low count categories in pie graph

        for row in self.channels:
            parts = row.split('\t')

            for category_name in categories_count_dict:
                if category_name == parts[0]:
                    categories_count_dict[category_name] += 1

        for category_name in categories_count_dict:
            if categories_count_dict[category_name] > 20:
                categories_name_list.append(category_name)
                categories_count_list.append(categories_count_dict[category_name])
            else:
                categories_count_dict['Others'] += categories_count_dict[category_name]

        frame = ctk.CTkFrame(
            self.main_window,
            fg_color="#100f0f",
            bg_color="#1C1B1A",
            corner_radius=12,
        )

        frame.place(x=260, y=120)

        ax.pie(
                categories_count_list,
                labels=categories_name_list,
                autopct='%1.1f%%',
                textprops={'color': '#FFFCF0'},
                shadow=True,
            )

        ax.set_title(title, color='#FFFCF0')
        fig.patch.set_facecolor('#100f0f') #whole
        ax.set_facecolor('#100f0f') # inside of the graph

        for category_name in categories_count_dict:
            if categories_count_dict[category_name] == max(categories_count_list):
                self.most_common_category_name = category_name

        # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
        canvas = FigureCanvasTkAgg(fig, master=frame)
        widget = canvas.get_tk_widget()
        widget.pack(padx=5, pady=5) # pad so rounded corners work

    def subgrowth_linegraph(self, y_part_value, x_part_value, title):
        plt.close('all') # now the graphs wont be on top of each other and multiple, but just one
        fig, ax = plt.subplots(figsize=(7.2, 5.6))

        selected_channels = self.channel_listbox.curselection()

        chosen_channels = []
        for index in selected_channels:
            chosen_channels.append(self.channel_listbox.get(index))

        for channel_name in chosen_channels:
            x_value = []
            y_value = []

            date_labels_first_count = 0
            date_labels_listcheck = []

            for row in self.timeseries:
                parts = row.split('\t')
                
                # Check just the ones selected in the ListBox
                if channel_name == self.channel_dict[parts[0]]:
                    y_value.append(float(parts[y_part_value]))
                    x_value.append(parts[x_part_value][:10]) # also remove hour and minute, [start:stop]

                    # in case of inconsistent amount of data, base the x graph axis of the first selected channel
                    if channel_name == chosen_channels[0]:
                        date_labels_first_count = len(x_value)
                        date_labels_listcheck = x_value


            if (len(x_value) == date_labels_first_count) and (x_value == date_labels_listcheck):
                ax.plot(x_value, y_value, label=channel_name)
                ax.set_xticks(x_value[::35]) # [start:stop:step]
            else:
                ax.plot(y_value, label=channel_name)

        frame = ctk.CTkFrame(
            self.main_window,
            fg_color="#100f0f",
            bg_color="#1C1B1A",
            corner_radius=12,
        )

        frame.place(x=510, y=120)

        self.style_ax(ax, fig)
        ax.legend(facecolor = '#100f0f')
        ax.grid(color='#343331')

        # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
        canvas = FigureCanvasTkAgg(fig, master=frame)
        widget = canvas.get_tk_widget()
        # fig.subplots_adjust(left = 0.05, right = 0.95, bottom = 0.05, top = 0.95) # from 0 to 1, from start to end of the window
        widget.pack(padx=5, pady=5) # pad so rounded corners work

    def channel_efficiency_linegraph(self, x_part_value, y_part_value, title):
        plt.close('all') # now the graphs wont be on top of each other and multiple, but just one
        fig, ax = plt.subplots(figsize=(7, 2.5))

        selected_channels = self.channel_listbox.curselection()

        chosen_channels = []
        for index in selected_channels:
            chosen_channels.append(self.channel_listbox.get(index))

        frame = ctk.CTkScrollableFrame(
            self.main_window,
            fg_color="#100f0f",
            bg_color="#1C1B1A",
            corner_radius=12,
            width=700,
            height=546
        )

        frame.place(x=510, y=120)

        for channel_name in chosen_channels:
            x_value = []
            y_value = []

            for row in self.timeseries:
                parts = row.split('\t')
                
                # Check just the ones selected in the ListBox
                if channel_name == self.channel_dict[parts[0]]:
                    y_value.append(float(parts[y_part_value]))
                    x_value.append(parts[x_part_value][:10]) # also remove hour and minute, [start:stop]


            fig, ax = plt.subplots(figsize=(7, 2.5))
            ax.plot(x_value, y_value, label=channel_name, color='#24837B')
            ax.set_xticks(x_value[::35]) # [start:stop:step]

            self.style_ax(ax, fig)
            ax.grid(color='#343331')
            # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
            canvas = FigureCanvasTkAgg(fig, master=frame)
            widget = canvas.get_tk_widget()
            # fig.subplots_adjust(left = 0.05, right = 0.95, bottom = 0.05, top = 0.95) # from 0 to 1, from start to end of the window
            widget.pack(padx=5, pady=5) # pad so rounded corners work

    def agevssuccess_scattergraph(self, y_part_value, x_part_value, title):
        selected_channels = self.channel_listbox.curselection()
        fig, ax = plt.subplots(figsize=(7.2, 5.6))

        # curselection gives int positions, need to convert to 'channel_name'
        chosen_channels = []
        for index in selected_channels:
            chosen_channels.append(self.channel_listbox.get(index))

        value_dict = {}
        value_dict_2 = {}
        x_value = []
        y_value = []
        for channel_name in chosen_channels:
            for row in self.channels:
                parts = row.split('\t')

                if parts[3] == channel_name:
                    value_dict[parts[x_part_value]] = parts[y_part_value]
                    value_dict_2[parts[x_part_value]] = channel_name

        value_dict = dict(sorted(value_dict.items()))
        value_dict_2 = dict(sorted(value_dict_2.items()))

        for key in value_dict:
            x_value = key
            y_value = (int(value_dict[key]))

            name = value_dict_2[key]

            ax.scatter(x_value, y_value, label=name)

        frame = ctk.CTkFrame(
            self.main_window,
            fg_color="#100f0f",
            bg_color="#1C1B1A",
            corner_radius=12,
        )

        frame.place(x=510, y=120)

        self.style_ax(ax, fig)
        ax.grid(color='#282726')
        ax.legend(facecolor = '#100f0f')
  
        # put the graph into canvas so it can be shown inside the customtikinter's frame inside its window
        canvas = FigureCanvasTkAgg(fig, master=frame)
        widget = canvas.get_tk_widget()
        # fig.subplots_adjust(left = 0.05, right = 0.95, bottom = 0.05, top = 0.95) # from 0 to 1, from start to end of the window
        widget.pack(padx=5, pady=5) # pad so rounded corners work

    def load_channel_dataset(self):
        self.channel_dict = {}
        self.channel_dict_join_date = {}
        with open('datasets/channels.tsv') as f:
            next(f)
            self.channels = f.read().strip().split('\n')

            for row in self.channels:
                parts = row.split('\t')

                self.channel_dict[parts[2]] = parts[3] # key = id, value = name
                self.channel_dict_join_date[parts[3]] = parts[1] # key = channel_name, value = join_date 

    def load_timeseries_dataset(self):
        with open('datasets/timeseries.tsv') as f:
            next(f)
            self.timeseries = f.read().strip().split('\n')

    def load_categories_listbox(self, frame):
        self.category_listbox = tk.Listbox(frame,
                     bg="#100f0f",
                     fg="#FFFCF0",
                     selectmode='single',
                     selectbackground="#282726",
                     selectforeground="#FFFCF0",
                     highlightthickness=0,
                     borderwidth=0,
                     width=20,
                     height=25,
                     font=('Courier New', 13))

        categories_list = []

        for row in self.channels:
            parts = row.split('\t')
            if parts[0] not in categories_list:
                categories_list.append(parts[0])
                self.category_listbox.insert(tk.END, parts[0])

        self.category_listbox.pack(padx=10, pady=10)

    def load_channel_listbox(self, frame):
        self.channel_listbox = tk.Listbox(frame,
                     bg="#100f0f",
                     fg="#FFFCF0",
                     selectmode='extended',
                     selectbackground="#282726",
                     selectforeground="#FFFCF0",
                     highlightthickness=0,
                     borderwidth=0,
                     width=20,
                     height=25,
                     font=('Courier New', 13))

        for row in self.channels:
            parts = row.split('\t')
            self.channel_listbox.insert(tk.END, parts[3])

        self.channel_listbox.pack(padx=10, pady=10)

    def letterform(self, n):
        n = float(n)
        if n >= 1000000000:
            return f"{n/1000000000:.1f}B"
        if n >= 1000000:
            return f"{n/1000000:.1f}M"
        if n >= 1000:
            return f"{n/1000:.1f}K"
        return str(int(n))

s = Data_analyzer()
s.load_channel_dataset()
s.load_timeseries_dataset()
s.main_menu()
s.overview_menu()
s.main_window.mainloop()