GameInsightify / module /__custom__.py
po5302006's picture
updated title format
acf3424
raw history blame
No virus
18.2 kB
########## LIBRARIES ##########
import streamlit as st
import pandas as pd
import numpy as np
import plotly.graph_objects as go
from streamlit_searchbox import st_searchbox
from module.__selectpage__ import st_page_selectbox
########## DATASET ##########
df = pd.read_csv('./data/join_02.csv')
df['date'] = pd.to_datetime(df['date']) # Format preparation
df['release_date'] = pd.to_datetime(df['release_date'])
df['avg_peak_perc'] = df['avg_peak_perc'].str.rstrip('%').astype('float')
df = df.dropna()
########## FUNCTION ##########
##### Adding single-player feature
def add_opp_features(genre):
df[genre[0]] = (df[genre[1]]==0)*1
##### Adding feature depending on range of a base feature
"""Scenario of using this function
For example, if we want a feature of price
between $10 to $30
"""
def add_range_features(arg):
lower = arg[0]; upper = arg[1]
name = arg[2]; genre = arg[3]
condition = (df[genre]>=lower) & (df[genre]<upper)
df[name] = condition*1
# Returning title version of feature name
def name(target):
return target.replace('_', ' ').title()
##### Searchbox Functions
# Appending favorite game selected by user to filtered list
"""use after search box
"""
def add_top_games(top_games, fav_games, ranges, df_ax):
ranges_last = ranges[1]
for fav in fav_games:
if fav in top_games:
range_last+=1
top_games = df_ax.gamename.unique()[ranges[0]-1:range_last]
elif fav!=None:
top_games = np.append(top_games, fav)
return top_games
def add_list(favorite_game, rec_games):
fav_games = []
if favorite_game[0]: fav_games = favorite_game
if rec_games:
fav_games = list(set(fav_games + rec_games))
return fav_games
# fav_games = []
# if favorite_game: fav_games = [favorite_game]
# if st.session_state.gamenames[-1]:
# rec_games = st.session_state.gamenames[-1]
# fav_games = list(set(fav_games + rec_games))
# def add_top_games(top_games, favorite_game, ranges, df_ax):
# if favorite_game in top_games:
# top_games = df_ax.gamename.unique()[ranges[0]-1:ranges[1]+1]
# elif favorite_game!=None:
# top_games = np.append(top_games, favorite_game)
# return top_games
# Linear search over all gamenames
def search(target):
gamenames = df['gamename'].unique() # all unique gamenames
result = []
for gamename in gamenames:
if target.lower() in gamename.lower(): # games that contains the searching keyword
result.append(gamename)
return result
# Streamlit search box
def searchbox():
selected_game = st_searchbox(
search_function=search,
key="gamename_searchbox",
default_options=None,
placeholder="Compare with your Favorite Game...",
)
return selected_game
# Overloaded with target name
def searchbox(target):
col1, col2= st.columns([1,1])
with col1:
selected_game = st_searchbox(
search_function=search,
key="gamename_searchbox",
default_options=target,
placeholder="Compare with your Favorite Game...",
)
return selected_game
########## PAGE SECTION ##########
# Datafram Section
"""Dataframe
together with Title
"""
def dfbox(ax_name, y_name, df_ax, ranges, order_name):
title = f"1.2 :blue[{ax_name}] Games with the {order_name} :blue[{y_name}]:"
with st.sidebar:
gamenames = df_ax.gamename.unique()
df_names = pd.DataFrame(gamenames, columns=['gamename'])
st.write(title)
st.dataframe(df_names[ranges[0]:ranges[1]+1])
def rec_dfbox():
title = f"1.1 :blue[Recommended] by :green[GameInsightify]"
if len(st.session_state.gamenames) > 0:
with st.sidebar:
rec_games = st.session_state.gamenames[-1]
df_names = pd.DataFrame(rec_games, columns=['gamename'])
st.write(title)
st.dataframe(df_names[0:len(rec_games)])
# Overloaded with argument of names
def home_dfbox(rec_games):
title = f":blue[Recommended] by :green[GameInsightify]"
if len(rec_games) > 0:
with st.sidebar:
df_names = pd.DataFrame(rec_games, columns=['gamename'])
st.write(title)
st.dataframe(df_names[0:len(rec_games)])
# plot 1 Section
"""Plot contains the top ranked games
based on the selected features,
within selected genre
"""
def plot1_box(ax, y, order_name, ranges, df_ax, top_games):
ax_name = name(ax) # formating strings
y_name = name(y) # formeting strings
title = f"1.3 Rank {ranges[0]} to {ranges[1]} :blue[{ax_name}] Games with the :red[{order_name}] :blue[{y_name}]"
st.subheader(title)
# Plot 1 - select box
rec_games = []
if len(st.session_state.gamenames) > 0 : rec_games = st.session_state.gamenames[-1]
favorite_game = searchbox(None) # search box to add a user favorite game on Plot 1
fav_games = add_list([favorite_game], rec_games)
fav_options = st.multiselect('Select Recommended Games', fav_games)
options = top_games
selected_tops = st.multiselect('Select Video Games', options)
selected_options = add_top_games(selected_tops, fav_options, ranges, df_ax)
# Plot 1
title_names = ','.join(selected_options)
plot_title = f"Monthly {y_name} of {title_names} Over Time"
gb = df.sort_values(by='date')
gb_list = {game: gb[gb["gamename"] == game] for game in selected_options}
fig_1 = go.Figure()
fig_1.update_layout(
title = plot_title,
xaxis_title = 'Date',
yaxis_title = y_name,
)
for game, gb in gb_list.items():
fig_1 = fig_1.add_trace(go.Scatter(x=gb["date"], y=gb[y], name=game, mode='lines'))
st.plotly_chart(fig_1)
def plot2_box(theme, y, genres, df_bx):
y_name = name(y)
title = f"2.0 Comparison Among :blue[{theme}] on Monthly :blue[{y_name}]:"
st.subheader(title)
# Plot 2 - Multiselect box
options = genres
selected_options = st.multiselect('Select Comparing Categories', options)
selected_names = ','.join(selected_options) # formating titles
plot_title = f"Monthly {y_name} of {selected_names} Over Time"
# Plot 2
# Tab 1 - Mean Line Plot
gb = df_bx.sort_values(by='date') # New copy of df
mean_list = {genre: gb[gb[genre] == 1].groupby('date').mean(y).reset_index() for genre in selected_options}
fig_mean = go.Figure()
for genre, gb in mean_list.items():
fig_mean = fig_mean.add_trace(go.Scatter(x=gb['date'], y=gb[y], name=genre, mode='lines'))
fig_mean.update_layout(
title = 'Mean of ' + plot_title,
xaxis_title = 'Date',
yaxis_title = 'Mean of '+y_name,
)
# Tab 2 - Sum Line Plot
gb = df_bx.sort_values(by='date')
sum_list = {genre: gb[gb[genre] == 1].groupby('date').sum(y).reset_index() for genre in selected_options}
fig_sum = go.Figure()
for genre, gb in sum_list.items():
fig_sum = fig_sum.add_trace(go.Scatter(x=gb['date'], y=gb[y], name=genre))
fig_sum.update_layout(
title = 'Sum of ' + plot_title,
xaxis_title='Date',
yaxis_title='Sum of '+y_name,
)
# Tab 3 - Scatter / Marker Plot
gb = df_bx.sort_values(by='date')
gb_list = {genre: gb[gb[genre] == 1] for genre in selected_options}
fig_sc = go.Figure()
for genre, gb in gb_list.items():
fig_sc = fig_sc.add_trace(go.Scatter(x=gb["date"], y=gb[y], name=genre, mode='markers'))
fig_sc.update_traces(
marker=dict(size=4, opacity=0.5)
)
fig_sc.update_layout(
title = plot_title,
xaxis_title='Date',
yaxis_title=y_name,
)
# Showing Plot
tab1, tab2, tab3 = st.tabs(['Line Plot', 'Sum Plot', 'Scatter Plot'])
with tab1:
st.plotly_chart(fig_mean)
with tab2:
st.plotly_chart(fig_sum)
with tab3:
st.plotly_chart(fig_sc)
# Plot 3 - Pie chart
import plotly.express as px
def plot3_box(theme, labels):
title = f"2.1 Ratio of Games Among :blue[{theme}]"
st.subheader(title)
if (type(labels)==str):
values = []
index = df[labels].unique().tolist()
for idx, value in enumerate(index):
count = len(df[df[labels] == value])
values.append(count)
if(count/len(df) < 0.02): index[idx] = 'Other'
df_p = pd.DataFrame(data = values,
index = index,
columns = ['counts'])
df_p = df_p.reset_index().rename(columns={'index':labels})
fig_ratio = px.pie(df_p, values='counts', names=labels)
st.plotly_chart(fig_ratio)
else:
values = []
for label in labels:
value = len(df[df[label]==1])
values.append(value)
fig_ratio = go.Figure(data=[go.Pie(labels=labels, values=values)])
st.plotly_chart(fig_ratio)
# Could not overload function, so renamed it
def plot3_box_limit(theme, labels, limit_perc):
title = f"2.1 Ratio of Games Among :blue[{theme}] over :blue[{limit_perc*100}%]"
st.subheader(title)
values = []
index = df[labels].unique().tolist()
for idx, value in enumerate(index):
count = len(df[df[labels] == value])
values.append(count)
if(count/len(df) < limit_perc): index[idx] = 'Other'
df_p = pd.DataFrame(data = values,
index = index,
columns = ['counts'])
df_p = df_p.reset_index().rename(columns={'index':labels})
fig_ratio = px.pie(df_p, values='counts', names=labels)
st.plotly_chart(fig_ratio)
def plot_chat_box(y, query_num, top_games):
y_name = name(y) # formeting strings
title = f"1.2 Comparison on The {query_num} Best Recommended Games on :blue[{y_name}]"
st.subheader(title)
# Plot 1 - select box # search box to add a user favorite game on Plot 1
options = top_games
selected_options = st.multiselect('Select Video Games', options)
# Plot 1
title_names = ','.join(selected_options)
plot_title = f"Monthly {y_name} of {title_names} Over Time"
gb = df.sort_values(by='date')
gb_list = {game: gb[gb["gamename"] == game] for game in selected_options}
fig_1 = go.Figure()
fig_1.update_layout(
title = plot_title,
xaxis_title = 'Date',
yaxis_title = y_name,
)
for game, gb in gb_list.items():
fig_1 = fig_1.add_trace(go.Scatter(x=gb["date"], y=gb[y], name=game, mode='lines'))
st.plotly_chart(fig_1)
##### Execute Page #####
def exec_page(emoji, theme, page_genres):
# Select Page
st_page_selectbox(theme)
# Header
st.header(emoji)
st.header(f"Customized Plot on :blue[{theme}]")
##### FILTER #####
# Featuer for both axis
features = ['avg', 'gain', 'peak', 'avg_peak_perc']
features += ['metacritic_score', 'positive', 'negative']
genres = page_genres
##################
# User Menu
order = st.toggle(label='Rank the Worst Games', value=False) # descending order toggle switch
left_col, right_col = st.columns(2) # Columns dividing
with left_col: y = st.selectbox("Select a Feature (y-axis)", features) # feature select box (y axis of Plots)
with right_col: ax = st.selectbox("Select a Genre (legend)", genres) # category select box (filtering game basse on genre)
order_name='Worst' if order else 'Highest' # string formating
y_name = name(y) # string of names that would be used on Plot title
ax_name = name(ax)
# Data - sorting and filtering
df_ax = df[df[ax]==1]
df_ax = df_ax[['gamename', 'date', y, ax]].sort_values(by=y, ascending=order).reset_index() # Data for Plot 1
df_bx = df[['gamename', 'date', y]+genres].sort_values(by=y, ascending=order).reset_index() # Data for Plot 2
# Slider
max = df_ax.gamename.unique().tolist() # max number of games
max = len(max)-1
ranges = st.slider(
label=f'Select range of the {order_name.lower()} games',
value = (1, 3),
min_value=1, max_value=30,
# min_value=1, max_value=max,
)
top_games = df_ax.gamename.unique()[ranges[0]-1:ranges[1]]
# Dataframe preview
rec_dfbox()
dfbox(ax_name, y_name, df_ax, ranges, order_name)
##### PLOT 1 #####
# Plot 1 - markdown
st.markdown("""***""")
plot1_box(ax, y, order_name, ranges, df_ax, top_games)
##### PLOT 2 #####
# Plot 2 - markdown
st.markdown("""***""")
plot2_box(theme, y, genres, df_bx)
##### HOME PAGE #####
def exec_page_home(theme):
st_page_selectbox(theme)
# Header
st.header("👋")
st.header("Customized Plot on :blue[General Features]")
##### FILTER #####
# Featuer for both axis
features = ['avg', 'gain', 'peak', 'avg_peak_perc']
genres = features
left_col, right_col = st.columns(2)
order = st.toggle(label='Rank the Worst Games', value=False) # descending order toggle switch
y = st.selectbox("Select a Feature (y-axis)", features) # feature select box
order_name='Worst' if order else 'Highest' # string formating
y_name = name(y)
# Data - sorting and filtering
df_ax = df[['gamename', 'date', y]].sort_values(by=y, ascending=order).reset_index() # Data - Plot 1
# df_bx = df[['gamename', 'date']+features].sort_values(by=y, ascending=order).reset_index() # Data - Plot 2
# Slider
max = df_ax.gamename.unique().tolist()
max = len(max)-1
ranges = st.slider(
label=f'Select range of the {order_name.lower()} games',
value = (1, 3),
min_value=1, max_value=30,
# min_value=1, max_value=max,
)
top_games = df_ax.gamename.unique()[ranges[0]-1:ranges[1]]
# Dataframe preview
rec_dfbox()
dfbox("", y_name, df_ax, ranges, order_name)
##### PLOT 1 #####
# Plot 1 - markdown
st.markdown("""***""")
title = f"1.3 Rank {ranges[0]} to {ranges[1]} Games with the Overall :red[{order_name}] :blue[{y_name}]"
st.subheader(title)
# Plot 1 - select box
rec_games = []
if len(st.session_state.gamenames)>0: rec_games = st.session_state.gamenames[-1]
favorite_game = searchbox(None) # search box to add a user favorite game on Plot 1
fav_games = add_list([favorite_game], rec_games)
fav_options = st.multiselect('Select Recommended Games', fav_games)
options = top_games
selected_tops = st.multiselect('Select Video Games', options)
selected_options = add_top_games(selected_tops, fav_options, ranges, df_ax)
# Plot 1
title_names = ','.join(selected_options)
plot_title = f"Monthly {y_name} of {title_names} Over Time"
gb = df_ax.sort_values(by='date')
gb_list = {game: gb[gb["gamename"] == game] for game in selected_options}
fig_1 = go.Figure()
fig_1.update_layout(
title = plot_title,
xaxis_title = 'Date',
yaxis_title = y_name,
)
for game, gb in gb_list.items():
fig_1 = fig_1.add_trace(go.Scatter(x=gb["date"], y=gb[y], name=game, mode='lines'))
st.plotly_chart(fig_1)
##### PUBLISHERS PAGE #####
def exec_page_pub(emoji, theme, main_genre):
st_page_selectbox(theme)
# Header
st.header(emoji)
st.header(f"Customized Plot on :blue[{theme}]")
##### FILTER #####
# Featuer for both axis
features = ['avg', 'gain', 'peak', 'avg_peak_perc']
features += ['metacritic_score', 'positive', 'negative']
genres = []
left_col, right_col = st.columns(2)
order = st.toggle(label='Find the Worst Games', value=False) # descending order toggle switch
with left_col:
y = st.selectbox("Select a Feature", features) # feature select box
with right_col:
if (main_genre=='publishers'):
genres = df.sort_values(by=y, ascending=order).publishers.unique()[0:5].tolist()
elif (main_genre=='developers'):
genres = df.sort_values(by=y, ascending=order).developers.unique()[0:5].tolist()
for genre in genres:
df[genre] = (df[main_genre]==genre)*1
ax = st.selectbox("Select a Category", genres) # category select box
order_name='Worst' if order else 'Highest' # string formating
y_name = y.replace('_', ' ').title()
ax_name = ax.title().replace('_', ' ')
# ### adding best publisher features feature ###
# Data - sorting and filtering
df_ax = df[df[ax]==1]
df_ax = df_ax[['gamename', 'date', y, ax]].sort_values(by=y, ascending=order).reset_index() # Data - Plot 1
df_bx = df[['gamename', 'date', y]+genres].sort_values(by=y, ascending=order).reset_index() # Data - Plot 2
# Slider
max = df_ax.gamename.unique().tolist()
max = len(max)
if(max < 2):value_r = 0
elif(max > 4):value_r = 5
else: value_r = max
ranges = st.slider(
label=f'Select range of the {order_name.lower()} games',
value = (1, value_r),
# min_value=0, max_value=30,
min_value=1, max_value=max,
)
top_games = df_ax.gamename.unique()[ranges[0]-1:ranges[1]]
# Dataframe preview
rec_dfbox()
dfbox(ax_name, y_name, df_ax, ranges, order_name)
title = f"1.2 5 :blue[{theme}s] with the :red[{order_name}] Monthly :blue[{y_name}]:"
st.subheader(title)
st.dataframe(genres[0:5])
##### PLOT 1 #####
# Plot 1 - markdown
st.markdown("""***""")
plot1_box(ax, y, order_name, ranges, df_ax, top_games)
##### PLOT 2 #####
# Plot 2 - markdown
st.markdown("""***""")
plot2_box(theme, y, genres, df_bx)