Visualization / app.py
TahaRasouli's picture
Rename visualization_tool.py to app.py
06a7dda verified
raw
history blame
16.7 kB
import streamlit as st
import pandas as pd
import plotly.express as px
import plotly.graph_objects as go
import numpy as np
def rgb_to_hex(r, g, b):
return f'#{r:02X}{g:02X}{b:02X}'
def interpolate_color(start_color, end_color, factor):
r1, g1, b1 = start_color
r2, g2, b2 = end_color
r = int(r1 + (r2 - r1) * factor)
g = int(g1 + (g2 - g1) * factor)
b = int(b1 + (b2 - b1) * factor)
return rgb_to_hex(r, g, b)
def get_warm_to_cold_colors(n):
hot = (124, 21, 77) # Hot color RGB
medium = (0, 66, 76) # Medium color RGB
cold = (197, 246, 235) # Cold color RGB
colors = []
for i in range(n):
if i < n // 2:
factor = i / (n // 2)
color = interpolate_color(hot, medium, factor)
else:
factor = (i - n // 2) / (n // 2)
color = interpolate_color(medium, cold, factor)
colors.append(color)
return colors
PREDEFINED_COLORS = {
'SC_Q_Origin': '#A7BCC6', #specified random color
'SC_Q_H_state':'#A7BCC6', #specified random color
'SC_Q_H_scope': '#A7BCC6', #specified random color
'H_sector': '#1A636B',
'SC_Q_H_sector': '#1A636B',
"H_companysize": '#A7BCC6', #specified random color
'H_revenue': '#49677B',
'H_employee': '#125F51',
'SC_Q_H_employee': '#125F51',
'I_importance': '#074057',
'IB_imp_weighted': '#5B8394',
'I_importance_fut': '#01626F',
'IB_imp_fut_weighted': '#39808B',
'I_invest_share': '#7B8D24',
'IB_invest_share_weighted': '#BCCFD6',
'I_invest_share_fut': '#16936D',
'IB_invest_share_fut_weighted': '#1D6073',
'I_eneffincrease_fut': '#007B86',
'H_energyuse': '#245B60',
'IB_energyuse_fut': '#587081',
'IB_energyuse_weighted': '#146153',
'IB_energyuse_fut_weighted': '#035263',
'H_energyuse_classes': '#66889A',
'SC_Q_H_energyuse_classes': '#66889A',
'SC_Q_S23_turnover_energycost': '#245B60', #specified random color
'H_energyintensity': '#186B77',
'Des_Gesamtumsatz': '#4A8A95',
'Prod_Erw': '#064B55',
'Prod_BDB': '#A7BCC6'
}
column_name_map = {
'H_sector': 'Sector',
'SC_Q_H_sector': 'Sector',
'SC_Q_Origin': 'Origin',
'SC_Q_H_state': 'State',
'SC_Q_H_scope': 'SC_Q_H_scope',
'H_revenue': 'Revenue',
'H_employee': 'Employees',
'SC_Q_H_employee': 'Employees',
"H_companysize": 'Company Size',
'I_importance': 'Importance',
'IB_imp_weighted': 'Weighted Importance',
'I_importance_fut': 'Future Importance',
'IB_imp_fut_weighted': 'Future Weighted Importance',
'I_invest_share': 'Investment Share (Past)',
'IB_invest_share_weighted': 'Weighted Investment Share (Past)',
'I_invest_share_fut': 'Investment Share (Future)',
'IB_invest_share_fut_weighted': 'Weighted Investment Share (Future)',
'I_eneffincrease_fut': 'Future Energy Efficiency Increase',
'H_energyuse': 'Energy Use',
'SC_Q_S23_turnover_energycost': 'Turn over Energy Cose', #must be reviewed
'IB_energyuse_fut': 'Future Energy Use',
'IB_energyuse_weighted': 'Weighted Energy Use',
'IB_energyuse_fut_weighted': 'Weighted Future Energy Use',
'H_energyuse_classes': 'Energy Use Classes',
"SC_Q_H_energyuse_classes": 'Energy Use Classes',
'H_energyintensity': 'Energy Intensity',
'Des_Gesamtumsatz': 'Total Revenue',
'Prod_Erw': 'Product Development',
'Prod_BDB': 'Product BDB',
}
def create_visualizations(df, selected_column):
# Line graph
fig_line = px.line(df, y=selected_column, title=f"Line Graph for {selected_column}")
line_color = PREDEFINED_COLORS.get(selected_column, 'black')
fig_line.update_traces(line=dict(color=line_color))
fig_line.update_layout(width=800, height=400)
st.plotly_chart(fig_line)
# Bar chart
fig_bar = px.bar(df, y=selected_column, title=f"Bar Chart for {selected_column}")
bar_color = PREDEFINED_COLORS.get(selected_column, 'blue')
fig_bar.update_traces(marker_color=bar_color)
fig_bar.update_layout(width=800, height=400)
st.plotly_chart(fig_bar)
# Pie chart (for categorical data or numerical data with few unique values)
if df[selected_column].dtype == 'object' or df[selected_column].nunique() < 10:
value_counts = df[selected_column].value_counts()
fig_pie = px.pie(names=value_counts.index, values=value_counts.values, title=f"Distribution of {selected_column}")
# Use predefined colors for the pie chart slices
pie_colors = [PREDEFINED_COLORS.get(c, 'grey') for c in value_counts.index]
fig_pie.update_traces(marker=dict(colors=pie_colors))
fig_pie.update_layout(width=800, height=400)
st.plotly_chart(fig_pie)
else:
st.write(f"Pie chart not applicable for {selected_column} due to high number of unique values.")
# Streamlit app
st.title('Fraunhofer Database')
# File uploader
uploaded_file = st.file_uploader("Choose a CSV file", type="csv")
if uploaded_file is not None:
# Load the data
df = pd.read_csv(uploaded_file)
# Get column names
columns = df.columns.tolist()
df_renamed = df.rename(columns=column_name_map)
# Sidebar for user input
st.sidebar.header('Options')
view_option = st.sidebar.radio('Select View Option',
['View All Data', 'Select Columns', 'Filter Data', 'Natural Language Query', 'Visualize Data', 'General Visualization'])
if view_option == 'View All Data':
st.write(df)
elif view_option == 'Select Columns':
selected_columns = st.sidebar.multiselect('Select Columns', columns)
if selected_columns:
st.write(df[selected_columns])
else:
st.write('Please select at least one column.')
elif view_option == 'Filter Data':
filter_column = st.sidebar.selectbox('Select Column to Filter', columns)
filter_operator = st.sidebar.selectbox('Select Operator', ['==', '>', '<', '>=', '<=', '!='])
filter_value = st.sidebar.text_input('Enter Filter Value')
if st.sidebar.button('Apply Filter'):
if filter_value:
column_type = df[filter_column].dtype
if column_type == 'int64':
filter_value = int(filter_value)
elif column_type == 'float64':
filter_value = float(filter_value)
# Apply filter
filtered_df = df.query(f"`{filter_column}` {filter_operator} @filter_value")
st.write(filtered_df)
else:
st.write('Please enter a filter value.')
elif view_option == 'Natural Language Query':
query = st.text_input('Enter your query (e.g., "Show me rows where age is greater than 30")')
if query:
column, operator, value = parse_query(query, df)
if column and operator and value is not None:
filtered_df = df.query(f"`{column}` {operator} @value")
st.write(filtered_df)
else:
st.write("Couldn't understand the query. Please try rephrasing.")
elif view_option == 'Visualize Data':
st.sidebar.subheader('Visualization Options')
chart_type = st.sidebar.selectbox('Select Chart Type', ['Bar Chart', 'Stacked Bar Chart', 'Line Graph', 'Pie Chart'])
# Sliders for figure size
width = st.sidebar.slider('Select Figure Width', 400, 1200, 800)
height = st.sidebar.slider('Select Figure Height', 300, 800, 600)
# Bar Chart Example
if chart_type == 'Bar Chart':
x_axis = st.sidebar.selectbox('Select X-axis', columns)
y_axis = st.sidebar.selectbox('Select Y-axis', columns)
if x_axis and y_axis:
fig = px.bar(df, x=x_axis, y=y_axis, title=f'{column_name_map.get(y_axis, "Undefined Y-axis")} by {x_axis}')
# Apply predefined colors
if y_axis in PREDEFINED_COLORS:
fig.update_traces(marker_color=PREDEFINED_COLORS[y_axis])
fig.update_layout(
title=f'{column_name_map.get(y_axis, "Undefined Y-axis")} by {column_name_map.get(x_axis, "Undefined X-axis")}',
xaxis_title=column_name_map.get(x_axis, "Undefined X-axis"),
yaxis_title=column_name_map.get(y_axis, "Undefined Y-axis"),
width=width,
height=height
)
st.plotly_chart(fig)
elif chart_type == 'Stacked Bar Chart':
x_axis = st.sidebar.selectbox('Select X-axis', columns)
y_axis = st.sidebar.selectbox('Select Y-axis', columns)
secondary_y_axis = st.sidebar.selectbox('Select Secondary Y-axis (for stacking)', columns)
if x_axis and y_axis and secondary_y_axis:
fig = go.Figure(data=[
go.Bar(name=f'{column_name_map.get(y_axis, "Undefined Y-axis")}', x=df[x_axis], y=df[y_axis], marker_color=PREDEFINED_COLORS.get(y_axis, 'blue')),
go.Bar(name=f'{column_name_map.get(secondary_y_axis, "Undefined Secondary Y-axis")}', x=df[x_axis], y=df[secondary_y_axis], marker_color=PREDEFINED_COLORS.get(secondary_y_axis, 'green'))
])
fig.update_layout(barmode='stack', title=f'{column_name_map.get(y_axis, "Undefined Y-axis")} and {column_name_map.get(secondary_y_axis, "Undefined Secondary Y-axis")} by {column_name_map.get(x_axis, "Undefined X-axis")}', width=width, height=height)
st.plotly_chart(fig)
elif chart_type == 'Line Graph':
x_axis = st.sidebar.selectbox('Select X-axis for Line Graph', columns)
y_axes = st.sidebar.multiselect('Select Y-axes for Line Graph', columns)
if x_axis and y_axes:
fig = go.Figure()
for y_axis in y_axes:
fig.add_trace(go.Scatter(
x=df[x_axis],
y=df[y_axis],
mode='lines',
name=column_name_map.get(y_axis, y_axis), # Use mapped name for the legend
line=dict(color=PREDEFINED_COLORS.get(y_axis, 'black'))
))
# Update layout with custom titles
fig.update_layout(
title=f'{", ".join(column_name_map.get(y_axis, y_axis) for y_axis in y_axes)} over {column_name_map.get(x_axis, x_axis)}',
xaxis_title=column_name_map.get(x_axis, "Undefined X-axis"),
yaxis_title="Values", # You can customize this or use a different mapping
width=width,
height=height
)
st.plotly_chart(fig)
# Pie Chart
elif chart_type == 'Pie Chart':
category_column = st.sidebar.selectbox('Select Category for Pie Chart', columns)
values_column = st.sidebar.selectbox('Select Values for Pie Chart', columns)
if category_column and values_column:
st.subheader(f'Pie Chart: {column_name_map.get(values_column, values_column)} by {column_name_map.get(category_column, category_column)}')
# Get the value counts
value_counts = df.groupby(category_column)[values_column].sum().sort_values(ascending=False)
# Generate warm to cold colors based on the number of slices
n_slices = len(value_counts)
color_map = get_warm_to_cold_colors(n_slices)
# Create the pie chart using the generated color map
fig = px.pie(df, names=value_counts.index, values=value_counts.values,
title=f'{column_name_map.get(values_column, values_column)} distribution by {column_name_map.get(category_column, category_column)}',
color_discrete_sequence=color_map)
fig.update_layout(width=width, height=height)
st.plotly_chart(fig)
elif view_option == 'General Visualization':
st.subheader('General Visualization')
# Function to get columns starting with a specific prefix
def get_columns_with_prefix(df, prefix):
return [col for col in df.columns if col.startswith(prefix)]
# Get columns for each category
h_columns = get_columns_with_prefix(df, 'H_')
i_columns = get_columns_with_prefix(df, 'I_')
ib_columns = get_columns_with_prefix(df, 'IB_')
# Create visualizations for a selected column
def create_visualizations(df, selected_column):
# Line graph
fig_line = px.line(df, y=selected_column,
title=f"Line Graph for {column_name_map.get(selected_column, selected_column)}")
line_color = PREDEFINED_COLORS.get(selected_column, 'black')
fig_line.update_traces(line=dict(color=line_color))
fig_line.update_layout(
xaxis_title=column_name_map.get('x_axis', 'Index'), # Optional: update X-axis title if relevant
yaxis_title=column_name_map.get(selected_column, 'Y-axis'), # Map Y-axis title
width=800, height=400
)
st.plotly_chart(fig_line)
# Bar chart
fig_bar = px.bar(df, y=selected_column,
title=f"Bar Chart for {column_name_map.get(selected_column, selected_column)}")
bar_color = PREDEFINED_COLORS.get(selected_column, 'blue')
fig_bar.update_traces(marker_color=bar_color)
fig_bar.update_layout(
xaxis_title=column_name_map.get('x_axis', 'Index'), # Optional: update X-axis title if relevant
yaxis_title=column_name_map.get(selected_column, 'Y-axis'), # Map Y-axis title
width=800, height=400
)
st.plotly_chart(fig_bar)
# Pie chart (for categorical data or numerical data with few unique values)
if df[selected_column].dtype == 'object' or df[selected_column].nunique() < 10:
value_counts = df[selected_column].value_counts()
fig_pie = px.pie(names=value_counts.index, values=value_counts.values,
title=f"Distribution of {column_name_map.get(selected_column, selected_column)}")
# Use predefined colors for the pie chart slices
pie_colors = [PREDEFINED_COLORS.get(c, 'grey') for c in value_counts.index]
fig_pie.update_traces(marker=dict(colors=pie_colors))
fig_pie.update_layout(width=800, height=400)
st.plotly_chart(fig_pie)
else:
st.write(f"Pie chart not applicable for {column_name_map.get(selected_column, selected_column)} due to high number of unique values.")
# Create tabs for different visualizations
tabs = st.tabs(["H Columns", "I Columns", "IB Columns"])
with tabs[0]:
st.subheader("H Columns Visualization")
h_selected = st.selectbox("Select an H column to visualize", h_columns)
if h_selected:
create_visualizations(df, h_selected)
with tabs[1]:
st.subheader("I Columns Visualization")
i_selected = st.selectbox("Select an I column to visualize", i_columns)
if i_selected:
create_visualizations(df, i_selected)
with tabs[2]:
st.subheader("IB Columns Visualization")
ib_selected = st.selectbox("Select an IB column to visualize", ib_columns)
if ib_selected:
create_visualizations(df, ib_selected)
# Summary statistics
st.subheader("Summary Statistics")
all_selected = h_columns + i_columns + ib_columns
if all_selected:
# Create a new DataFrame for summary statistics with mapped column names
summary_df = df[all_selected].describe().T # Transpose for better readability
summary_df.index = [column_name_map.get(col, col) for col in summary_df.index] # Update index names
st.write(summary_df)
else:
st.write('Please upload your data.')
st.sidebar.info('Designed by Taha Rasouli at Fraunhofer')