VotoMatcher / app.py
Thiloid's picture
Update app.py
811d734 verified
Raw
History Blame Contribute Delete
12.3 kB
import streamlit as st
import json
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
footer="""<style>
a:link , a:visited{
color: blue;
background-color: transparent;
text-decoration: underline;
}
a:hover, a:active {
color: red;
background-color: transparent;
text-decoration: underline;
}
.footer {
position: fixed;
left: 0;
bottom: 0;
width: 100%;
background-color: cornflowerblue;
color: black;
text-align: center;
}
</style>
<div class="footer">
<p>This app was developed by <a href="https://www.voto.vote" target="_blank">VOTO</a>.
Developer: <a href="https://linkedin.com/in/thilo-dieing-bba5b2253" target="_blank">Thilo I. Dieing, M.Sc.</a></p>
</div>
"""
st.markdown(footer,unsafe_allow_html=True)
def add_logo():
st.markdown(
"""
<style>
[data-testid="stSidebarNav"]::before {
content: "Uni Mannheim ASR Team Project";
margin-left: 20px;
margin-top: 20px;
font-size: 20px;
position: relative;
top: 100px;
}
</style>
""",
unsafe_allow_html=True,
)
# functions
add_logo()
col1, col2 = st.columns(2)
with col1:
st.image("voto_purple.png", width=300)
with col2:
st.image("tuda_logo.tif", width=300)
# Title of the app
st.title("VOTO party insights")
# Allow the user to upload a file
uploaded_file = st.file_uploader("Upload the JSON file with your VOTO party answers", type=["json"])
# Check if a file is uploaded
if uploaded_file is not None:
try:
# Try to load the uploaded file as JSON
file_content = uploaded_file.read()
json_data = json.loads(file_content)
# If the file is valid JSON, show a success message
# Convert the JSON data to a DataFrame
if isinstance(json_data, list): # Expecting a list of dictionaries for a proper DataFrame
df = pd.DataFrame(json_data)
election=str(df["instance"].iloc[0])
statements = str(df["statement"].nunique())
statement_n = df["statement"].nunique()
try:
parties = str(df["party_name"].nunique())
except KeyError:
parties = str(df["party"].nunique())
df["party_name"] = df["party"]
st.success("You have uploaded the party answers for the VOTO instance: "+election+" ("+statements+" statements; "+parties+" Parties)")
expander1 = st.expander("Statement Selection")
pivot_df = df.pivot(index='statement', columns='party_name', values='valuation')
# std
df['std_deviation'] = df.groupby('statement')['valuation'].transform('std')
# how many differnet postions
df['unique_postions'] = df.groupby('statement')['valuation'].transform('nunique')
# how many neutrals
df['n_neutral'] = df.groupby('statement')['valuation'].transform(lambda x: (x == 50).sum())
# standalone party postions
value_counts = df.groupby(['statement', 'valuation']).size().reset_index(name='Count')
unique_values = value_counts[value_counts['Count'] == 1]
single_counts = unique_values.groupby('statement').size().reset_index(name='SingleCount')
df = df.merge(single_counts, on='statement', how='left')
df['SingleCount'] = df['SingleCount'].fillna(0).astype(int)
#std without neutral
df_filtered = df[df['valuation'] != 50]
df_filtered['std_deviation2'] = df_filtered.groupby('statement')['valuation'].transform('std')
df2= df_filtered[["statement",'std_deviation2' ]].drop_duplicates()
df = df.merge(df2, on='statement', how='left', suffixes=('', '_new'))
sorted_single_counts = df[['statement', 'std_deviation']].drop_duplicates()
sorted_single_counts['AdjustedRank1'] = sorted_single_counts['std_deviation'].rank(method='dense', ascending=False).astype(int) - 1
df = df.merge(sorted_single_counts[['statement', 'AdjustedRank1']], on='statement', how='left', suffixes=('', '_new'))
sorted_single_counts = df[['statement', 'unique_postions']].drop_duplicates()
sorted_single_counts['AdjustedRank2'] = sorted_single_counts['unique_postions'].rank(method='dense', ascending=False).astype(int) - 1
df = df.merge(sorted_single_counts[['statement', 'AdjustedRank2']], on='statement', how='left', suffixes=('', '_new'))
sorted_single_counts = df[['statement', 'n_neutral']].drop_duplicates()
sorted_single_counts['AdjustedRank3'] = sorted_single_counts['n_neutral'].rank(method='dense').astype(int) - 1
df = df.merge(sorted_single_counts[['statement', 'AdjustedRank3']], on='statement', how='left', suffixes=('', '_new'))
sorted_single_counts = df[['statement', 'SingleCount']].drop_duplicates()
sorted_single_counts['AdjustedRank4'] = sorted_single_counts['SingleCount'].rank(method='dense', ascending=False).astype(int) - 1
df = df.merge(sorted_single_counts[['statement', 'AdjustedRank4']], on='statement', how='left', suffixes=('', '_new'))
sorted_single_counts = df[['statement', 'std_deviation2']].drop_duplicates()
sorted_single_counts['AdjustedRank5'] = sorted_single_counts['std_deviation2'].rank(method='dense', ascending=False).astype(int) - 1
df = df.merge(sorted_single_counts[['statement', 'AdjustedRank5']], on='statement', how='left', suffixes=('', '_new'))
df["statement_importance"]= (df["AdjustedRank1"]+df["AdjustedRank2"]+df["AdjustedRank3"]+df["AdjustedRank4"]+df["AdjustedRank5"])/5
df_pres= df[['statement',"statement_importance"]].drop_duplicates()
df_sorted1 = df_pres.sort_values(by='statement_importance')
df_sorted1.reset_index(drop=True, inplace=True)
df_sorted = df.sort_values(by='statement_importance')
df_sorted.reset_index(drop=True, inplace=True)
statn = expander1.slider("How many statements do you want in your final VOTO? You currently have **"+statements+ "** statements." , 0, int(statements),0)
if statn !=0:
expander1.write("**Based on that our metric recommends keeping the following " +str(statn) +" statements:**")
lstat=df_sorted1['statement'].tolist()
lstat1=lstat[:statn]
i=0
for l in lstat1:
i=i+1
expander1.markdown("<span style='color: green;'>"+str(i)+": "+l+"</span>", unsafe_allow_html=True)
expander1.write("**While the metric recommends dropping "+str(statement_n-statn) +" statements due to a lack of difference:**")
lstat=df_sorted1['statement'].tolist()
lstat1=lstat[statn:]
i=statn
for l in lstat1:
i=i+1
expander1.markdown("<span style='color: red;'>"+str(i)+": "+l+"</span>", unsafe_allow_html=True)
if expander1.button("Additional information on individual metric scores"):
expander1.dataframe(df_sorted)
expander2 = st.expander("Party Positions")
df = pd.DataFrame(json_data)
try:
parties = str(df["party_name"].nunique())
except KeyError:
parties = str(df["party"].nunique())
df["party_name"] = df["party"]
unique_party_names = df['party_name'].unique()
unique_party_names = [""] + list(unique_party_names) # Add an empty option
selected_party = expander2.selectbox("Select a Party", unique_party_names)
contains_25_or_75 = (25 in df['valuation'].values) or (75 in df['valuation'].values)
if contains_25_or_75:
y_ticks = [0, 25, 50, 75, 100] # Set y-ticks for scale 1 to 5
y_tick_labels = ['Strong Disagreement', 'Disagreement', 'Neutral', 'Agreement', 'Strong Agreement'] # Custom labels
else:
y_ticks = [0, 50, 100] # Set y-ticks for scale 1 to 3
y_tick_labels = ['Disagreement', 'Neutral', 'Agreement']
if selected_party: # Check if a party has been selected
df['short_text'] = df['statement'].apply(lambda x: ' '.join(x.split()[:11]))
filtered_df = df[df['party_name'] == selected_party]
fig, ax = plt.subplots(figsize=(6, 20))
ax.scatter(filtered_df['valuation'], filtered_df['short_text'], color='purple', s=100) # s is the size of points
ax.set_xlabel('Valuation',fontsize=16)
ax.set_ylabel('Statements',fontsize=16)
ax.set_title('Valuation by Statement',fontsize=16)
ax.set_xticks(y_ticks) # Set y-ticks dynamically based on condition
ax.tick_params(axis='y', labelsize=16)
ax.tick_params(axis='x', labelsize=16)
ax.set_xticklabels(y_tick_labels)
plt.xticks(rotation=45)
ax.grid(True)
expander2.pyplot(fig)
expander3 = st.expander("Unique Party Positions")
df = pd.DataFrame(json_data)
try:
parties = str(df["party_name"].nunique())
except KeyError:
parties = str(df["party"].nunique())
df["party_name"] = df["party"]
unique_party_names = df['party_name'].unique()
unique_party_names = [""] + list(unique_party_names) # Add an empty option
selected_party = expander3.selectbox("Select a Party", unique_party_names, key="2")
contains_25_or_75 = (25 in df['valuation'].values) or (75 in df['valuation'].values)
if contains_25_or_75:
y_ticks = [0, 25, 50, 75, 100] # Set y-ticks for scale 1 to 5
y_tick_labels = ['Strong Disagreement', 'Disagreement', 'Neutral', 'Agreement', 'Strong Agreement'] # Custom labels
else:
y_ticks = [0, 50, 100] # Set y-ticks for scale 1 to 3
y_tick_labels = ['Disagreement', 'Neutral', 'Agreement']
if selected_party: # Check if a party has been selected
value_counts = df.groupby(['statement', 'valuation']).size().reset_index(name='Count')
unique_values = value_counts[value_counts['Count'] == 1]
unique_parties = df.merge(unique_values[['statement', 'valuation']], on=['statement', 'valuation'])
unique_parties_result = unique_parties[['statement', 'party_name', 'valuation']]
unique_parties_result['short_text'] = unique_parties_result['statement'].apply(lambda x: ' '.join(x.split()[:11]))
filtered_unique_parties_result = unique_parties_result[unique_parties_result['party_name'] == selected_party]
fig, ax = plt.subplots(figsize=(6, 12))
ax.scatter(filtered_unique_parties_result['valuation'], filtered_unique_parties_result['short_text'], color='purple', s=100) # s is the size of points
ax.set_xlabel('Valuation',fontsize=16)
ax.set_ylabel('Statements',fontsize=16)
ax.set_title('Valuation by Statement',fontsize=16)
ax.set_xticks(y_ticks) # Set y-ticks dynamically based on condition
ax.tick_params(axis='y', labelsize=16)
ax.tick_params(axis='x', labelsize=16)
ax.set_xticklabels(y_tick_labels)
plt.xticks(rotation=45)
ax.grid(True)
expander3.pyplot(fig)
else:
st.warning("The JSON file structure is not suitable for DataFrame conversion. It should be a list of dictionaries.")
except json.JSONDecodeError:
# If the file is not a valid JSON, show an error message
st.error("This is not a valid JSON file. Please upload a valid JSON.")
else:
st.info("Please upload a JSON file to get started.")