Add Gradio app and update dependencies
Browse files- .gitignore +4 -1
- app.py +260 -0
- notebooks/financial_sentiment_analysis.ipynb +0 -0
- requirements.txt +2 -1
.gitignore
CHANGED
@@ -18,4 +18,7 @@ data/processed/*
|
|
18 |
# Virtual environment
|
19 |
venv/
|
20 |
env/
|
21 |
-
.venv/
|
|
|
|
|
|
|
|
18 |
# Virtual environment
|
19 |
venv/
|
20 |
env/
|
21 |
+
.venv/
|
22 |
+
|
23 |
+
# Temporary files
|
24 |
+
temp_merged_data.csv
|
app.py
ADDED
@@ -0,0 +1,260 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import pandas as pd
|
3 |
+
from datetime import datetime, timedelta
|
4 |
+
import matplotlib.pyplot as plt
|
5 |
+
import matplotlib.dates as mdates
|
6 |
+
import os
|
7 |
+
import sys
|
8 |
+
import io
|
9 |
+
import base64
|
10 |
+
|
11 |
+
# Add src directory to path to import modules
|
12 |
+
module_path = os.path.abspath(os.path.join('.'))
|
13 |
+
if module_path not in sys.path:
|
14 |
+
sys.path.append(module_path)
|
15 |
+
|
16 |
+
# Import functions from your src directory
|
17 |
+
try:
|
18 |
+
from src.data_fetcher import get_stock_data, get_news_articles, load_api_keys
|
19 |
+
from src.sentiment_analyzer import analyze_sentiment
|
20 |
+
except ImportError as e:
|
21 |
+
# Handle error gracefully if run from a different directory or modules missing
|
22 |
+
print(f"Error importing modules from src: {e}. Ensure app.py is in the project root and src/* exists.")
|
23 |
+
# Define dummy functions if imports fail, so Gradio interface can still load
|
24 |
+
def get_stock_data(*args, **kwargs): return None
|
25 |
+
def get_news_articles(*args, **kwargs): return None
|
26 |
+
def analyze_sentiment(*args, **kwargs): return None, None, None
|
27 |
+
def load_api_keys(): return None, None
|
28 |
+
|
29 |
+
|
30 |
+
# --- Data Fetching and Processing Logic ---
|
31 |
+
# (Similar to the Streamlit version, but adapted for Gradio outputs)
|
32 |
+
def perform_analysis(ticker_symbol, start_date_str, end_date_str): # Renamed date inputs
|
33 |
+
"""Fetches data, analyzes sentiment, merges, and prepares outputs for Gradio."""
|
34 |
+
if not ticker_symbol:
|
35 |
+
return None, "Please enter a stock ticker.", None, None, None
|
36 |
+
|
37 |
+
# Ensure API keys are loaded (needed for news)
|
38 |
+
news_key, _ = load_api_keys()
|
39 |
+
if not news_key:
|
40 |
+
return None, "Error: NEWS_API_KEY not found in .env file. Cannot fetch news.", None, None, None
|
41 |
+
|
42 |
+
# Validate and parse date strings
|
43 |
+
try:
|
44 |
+
start_date_obj = datetime.strptime(start_date_str, '%Y-%m-%d').date()
|
45 |
+
end_date_obj = datetime.strptime(end_date_str, '%Y-%m-%d').date()
|
46 |
+
except ValueError:
|
47 |
+
return None, "Error: Invalid date format. Please use YYYY-MM-DD.", None, None, None
|
48 |
+
|
49 |
+
|
50 |
+
if start_date_obj >= end_date_obj:
|
51 |
+
return None, "Error: Start date must be before end date.", None, None, None
|
52 |
+
|
53 |
+
status_updates = f"Fetching data for {ticker_symbol} from {start_date_str} to {end_date_str}...\n"
|
54 |
+
|
55 |
+
# 1. Fetch Stock Data
|
56 |
+
stock_df = get_stock_data(ticker_symbol, start_date_str, end_date_str)
|
57 |
+
if stock_df is None or stock_df.empty:
|
58 |
+
status_updates += "Could not fetch stock data.\n"
|
59 |
+
# Return early if essential data is missing
|
60 |
+
return None, status_updates, None, None, None
|
61 |
+
else:
|
62 |
+
status_updates += f"Successfully fetched {len(stock_df)} days of stock data.\n"
|
63 |
+
stock_df['Date'] = pd.to_datetime(stock_df['Date'])
|
64 |
+
|
65 |
+
|
66 |
+
# 2. Fetch News Articles
|
67 |
+
articles_list = get_news_articles(ticker_symbol, start_date_str, end_date_str)
|
68 |
+
if articles_list is None or not articles_list:
|
69 |
+
status_updates += "Could not fetch news articles or none found.\n"
|
70 |
+
news_df = pd.DataFrame()
|
71 |
+
else:
|
72 |
+
status_updates += f"Found {len(articles_list)} potential news articles.\n"
|
73 |
+
news_df = pd.DataFrame(articles_list)
|
74 |
+
if 'publishedAt' in news_df.columns:
|
75 |
+
news_df['publishedAt'] = pd.to_datetime(news_df['publishedAt'])
|
76 |
+
news_df['date'] = news_df['publishedAt'].dt.date
|
77 |
+
news_df['date'] = pd.to_datetime(news_df['date']) # Convert date to datetime for merging
|
78 |
+
else:
|
79 |
+
status_updates += "Warning: News articles missing 'publishedAt' field.\n"
|
80 |
+
news_df['date'] = None
|
81 |
+
|
82 |
+
# 3. Sentiment Analysis (if news available)
|
83 |
+
daily_sentiment = pd.DataFrame(columns=['date', 'avg_sentiment_score']) # Default empty
|
84 |
+
if not news_df.empty and 'date' in news_df.columns and news_df['date'].notna().any():
|
85 |
+
status_updates += f"Performing sentiment analysis on {len(news_df)} articles...\n"
|
86 |
+
news_df['text_to_analyze'] = news_df['title'].fillna('') + ". " + news_df['description'].fillna('')
|
87 |
+
# --- Apply sentiment analysis ---
|
88 |
+
# This can be slow, consider progress updates if possible or running async
|
89 |
+
sentiment_results = news_df['text_to_analyze'].apply(lambda x: analyze_sentiment(x) if pd.notna(x) else (None, None, None))
|
90 |
+
news_df['sentiment_label'] = sentiment_results.apply(lambda x: x[0])
|
91 |
+
news_df['sentiment_score'] = sentiment_results.apply(lambda x: x[1])
|
92 |
+
status_updates += "Sentiment analysis complete.\n"
|
93 |
+
|
94 |
+
# 4. Aggregate Sentiment
|
95 |
+
valid_sentiment_df = news_df.dropna(subset=['sentiment_score', 'date'])
|
96 |
+
if not valid_sentiment_df.empty:
|
97 |
+
daily_sentiment = valid_sentiment_df.groupby('date')['sentiment_score'].mean().reset_index()
|
98 |
+
daily_sentiment.rename(columns={'sentiment_score': 'avg_sentiment_score'}, inplace=True)
|
99 |
+
status_updates += "Aggregated daily sentiment scores.\n"
|
100 |
+
else:
|
101 |
+
status_updates += "No valid sentiment scores found to aggregate.\n"
|
102 |
+
|
103 |
+
# 5. Merge Data
|
104 |
+
if not daily_sentiment.empty:
|
105 |
+
merged_df = pd.merge(stock_df, daily_sentiment, left_on='Date', right_on='date', how='left')
|
106 |
+
if 'date' in merged_df.columns:
|
107 |
+
merged_df.drop(columns=['date'], inplace=True)
|
108 |
+
status_updates += "Merged stock data with sentiment scores.\n"
|
109 |
+
else:
|
110 |
+
merged_df = stock_df.copy() # Keep stock data even if no sentiment
|
111 |
+
merged_df['avg_sentiment_score'] = None # Add column with None
|
112 |
+
status_updates += "No sentiment data to merge.\n"
|
113 |
+
|
114 |
+
# 6. Calculate Price Change and Lagged Sentiment for Correlation
|
115 |
+
merged_df['price_pct_change'] = merged_df['Close'].pct_change()
|
116 |
+
merged_df['sentiment_lagged'] = merged_df['avg_sentiment_score'].shift(1)
|
117 |
+
|
118 |
+
# --- Generate Outputs ---
|
119 |
+
|
120 |
+
# Plot
|
121 |
+
plot_object = None
|
122 |
+
if not merged_df.empty:
|
123 |
+
fig, ax1 = plt.subplots(figsize=(12, 6)) # Adjusted size for Gradio
|
124 |
+
|
125 |
+
color = 'tab:blue'
|
126 |
+
ax1.set_xlabel('Date')
|
127 |
+
ax1.set_ylabel('Stock Close Price', color=color)
|
128 |
+
ax1.plot(merged_df['Date'], merged_df['Close'], color=color, label='Stock Price')
|
129 |
+
ax1.tick_params(axis='y', labelcolor=color)
|
130 |
+
ax1.tick_params(axis='x', rotation=45)
|
131 |
+
ax1.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
|
132 |
+
ax1.xaxis.set_major_locator(mdates.AutoDateLocator(maxticks=10)) # Auto ticks
|
133 |
+
|
134 |
+
if 'avg_sentiment_score' in merged_df.columns and merged_df['avg_sentiment_score'].notna().any():
|
135 |
+
ax2 = ax1.twinx()
|
136 |
+
color = 'tab:red'
|
137 |
+
ax2.set_ylabel('Average Sentiment Score', color=color)
|
138 |
+
ax2.plot(merged_df['Date'], merged_df['avg_sentiment_score'], color=color, linestyle='--', marker='o', markersize=4, label='Avg Sentiment')
|
139 |
+
ax2.tick_params(axis='y', labelcolor=color)
|
140 |
+
ax2.axhline(0, color='grey', linestyle='--', linewidth=0.8)
|
141 |
+
ax2.set_ylim(-1.1, 1.1) # Fix sentiment axis range
|
142 |
+
|
143 |
+
# Combine legends
|
144 |
+
lines, labels = ax1.get_legend_handles_labels()
|
145 |
+
lines2, labels2 = ax2.get_legend_handles_labels()
|
146 |
+
ax2.legend(lines + lines2, labels + labels2, loc='upper left')
|
147 |
+
else:
|
148 |
+
ax1.legend(loc='upper left') # Only stock legend
|
149 |
+
|
150 |
+
plt.title(f"{ticker_symbol} Stock Price vs. Average Daily News Sentiment")
|
151 |
+
plt.grid(True, which='major', linestyle='--', linewidth='0.5', color='grey')
|
152 |
+
fig.tight_layout()
|
153 |
+
plot_object = fig # Return the figure object for Gradio plot component
|
154 |
+
status_updates += "Generated plot.\n"
|
155 |
+
|
156 |
+
|
157 |
+
# Correlation & Insights Text
|
158 |
+
insights_text = "## Analysis Results\n\n"
|
159 |
+
correlation = None
|
160 |
+
if 'sentiment_lagged' in merged_df.columns and merged_df['sentiment_lagged'].notna().any() and merged_df['price_pct_change'].notna().any():
|
161 |
+
correlation_df = merged_df[['sentiment_lagged', 'price_pct_change']].dropna()
|
162 |
+
if not correlation_df.empty and len(correlation_df) > 1:
|
163 |
+
correlation = correlation_df['sentiment_lagged'].corr(correlation_df['price_pct_change'])
|
164 |
+
insights_text += f"**Correlation (Lagged Sentiment vs Price Change):** {correlation:.4f}\n"
|
165 |
+
insights_text += "_Measures correlation between the previous day's average sentiment and the current day's price percentage change._\n\n"
|
166 |
+
else:
|
167 |
+
insights_text += "Correlation: Not enough overlapping data points to calculate.\n\n"
|
168 |
+
else:
|
169 |
+
insights_text += "Correlation: Sentiment or price change data missing.\n\n"
|
170 |
+
|
171 |
+
# Simple Insights
|
172 |
+
insights_text += "**Potential Insights (Not Financial Advice):**\n"
|
173 |
+
if 'avg_sentiment_score' in merged_df.columns and merged_df['avg_sentiment_score'].notna().any():
|
174 |
+
avg_sentiment_overall = merged_df['avg_sentiment_score'].mean()
|
175 |
+
insights_text += f"- Average Sentiment (Overall Period): {avg_sentiment_overall:.3f}\n"
|
176 |
+
|
177 |
+
if correlation is not None and pd.notna(correlation):
|
178 |
+
if correlation > 0.15:
|
179 |
+
insights_text += "- Positive correlation detected. Higher sentiment yesterday tended to correlate with price increases today.\n"
|
180 |
+
elif correlation < -0.15:
|
181 |
+
insights_text += "- Negative correlation detected. Higher sentiment yesterday tended to correlate with price decreases today (or vice-versa).\n"
|
182 |
+
else:
|
183 |
+
insights_text += "- Weak correlation detected. Sentiment may not be a strong short-term driver for this period.\n"
|
184 |
+
else:
|
185 |
+
insights_text += "- No sentiment data available to generate insights.\n"
|
186 |
+
|
187 |
+
insights_text += "\n**Disclaimer:** This analysis is automated and NOT financial advice. Many factors influence stock prices."
|
188 |
+
status_updates += "Generated insights.\n"
|
189 |
+
|
190 |
+
# Recent News DataFrame
|
191 |
+
recent_news_df = pd.DataFrame()
|
192 |
+
if not news_df.empty and 'publishedAt' in news_df.columns:
|
193 |
+
# Select and format columns for display
|
194 |
+
cols_to_show = ['publishedAt', 'title', 'sentiment_label', 'sentiment_score']
|
195 |
+
# Ensure all columns exist before selecting
|
196 |
+
cols_exist = [col for col in cols_to_show if col in news_df.columns]
|
197 |
+
if cols_exist:
|
198 |
+
recent_news_df = news_df.sort_values(by='publishedAt', ascending=False)[cols_exist].head(10)
|
199 |
+
# Format date for display
|
200 |
+
recent_news_df['publishedAt'] = recent_news_df['publishedAt'].dt.strftime('%Y-%m-%d %H:%M')
|
201 |
+
status_updates += "Prepared recent news table.\n"
|
202 |
+
|
203 |
+
|
204 |
+
return plot_object, insights_text, recent_news_df, status_updates, merged_df # Return merged_df for potential download
|
205 |
+
|
206 |
+
|
207 |
+
# --- Gradio Interface Definition ---
|
208 |
+
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
209 |
+
gr.Markdown("# Stock Sentiment Analysis Dashboard")
|
210 |
+
|
211 |
+
with gr.Row():
|
212 |
+
with gr.Column(scale=1):
|
213 |
+
ticker_input = gr.Textbox(label="Stock Ticker", value="AAPL", placeholder="e.g., AAPL, GOOGL")
|
214 |
+
# Use Textbox for dates, value should be string
|
215 |
+
start_date_input = gr.Textbox(label="Start Date (YYYY-MM-DD)", value=(datetime.now() - timedelta(days=30)).strftime('%Y-%m-%d'))
|
216 |
+
end_date_input = gr.Textbox(label="End Date (YYYY-MM-DD)", value=datetime.now().strftime('%Y-%m-%d'))
|
217 |
+
analyze_button = gr.Button("Analyze", variant="primary")
|
218 |
+
status_output = gr.Textbox(label="Analysis Status", lines=5, interactive=False)
|
219 |
+
# Optional: Add download button for the merged data
|
220 |
+
download_data = gr.File(label="Download Merged Data (CSV)")
|
221 |
+
|
222 |
+
|
223 |
+
with gr.Column(scale=3):
|
224 |
+
plot_output = gr.Plot(label="Stock Price vs. Sentiment")
|
225 |
+
insights_output = gr.Markdown(label="Analysis & Insights")
|
226 |
+
news_output = gr.DataFrame(label="Recent News Headlines", headers=['Date', 'Title', 'Sentiment', 'Score'], wrap=True)
|
227 |
+
|
228 |
+
# Hidden state to store the merged dataframe for download
|
229 |
+
merged_df_state = gr.State(None)
|
230 |
+
|
231 |
+
def run_analysis_and_prepare_download(ticker, start_date_str, end_date_str): # Use string names
|
232 |
+
"""Wrapper function to run analysis and prepare CSV for download."""
|
233 |
+
# Parse dates inside the wrapper or ensure perform_analysis handles strings robustly
|
234 |
+
try:
|
235 |
+
start_date_obj = datetime.strptime(start_date_str, '%Y-%m-%d').date()
|
236 |
+
end_date_obj = datetime.strptime(end_date_str, '%Y-%m-%d').date()
|
237 |
+
except ValueError:
|
238 |
+
# Handle invalid date format input from textbox
|
239 |
+
return None, "Error: Invalid date format. Please use YYYY-MM-DD.", None, "Error processing dates.", None, None
|
240 |
+
|
241 |
+
plot, insights, news, status, merged_df = perform_analysis(ticker, start_date_str, end_date_str) # Pass strings
|
242 |
+
|
243 |
+
csv_path = None
|
244 |
+
if merged_df is not None and not merged_df.empty:
|
245 |
+
# Save to a temporary CSV file for Gradio download
|
246 |
+
csv_path = "temp_merged_data.csv"
|
247 |
+
merged_df.to_csv(csv_path, index=False)
|
248 |
+
|
249 |
+
return plot, insights, news, status, merged_df, csv_path # Return path for download
|
250 |
+
|
251 |
+
|
252 |
+
analyze_button.click(
|
253 |
+
fn=run_analysis_and_prepare_download,
|
254 |
+
inputs=[ticker_input, start_date_input, end_date_input], # Inputs are now textboxes
|
255 |
+
outputs=[plot_output, insights_output, news_output, status_output, merged_df_state, download_data] # Update state and file output
|
256 |
+
)
|
257 |
+
|
258 |
+
# --- Launch the App ---
|
259 |
+
if __name__ == "__main__":
|
260 |
+
demo.launch()
|
notebooks/financial_sentiment_analysis.ipynb
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
requirements.txt
CHANGED
@@ -7,4 +7,5 @@ transformers
|
|
7 |
scikit-learn
|
8 |
matplotlib
|
9 |
nltk
|
10 |
-
python-dotenv
|
|
|
|
7 |
scikit-learn
|
8 |
matplotlib
|
9 |
nltk
|
10 |
+
python-dotenv
|
11 |
+
gradio
|