Update app.py

app.py

@@ -5,11 +5,9 @@ import plotly.express as px
 import yfinance as yf
 import gradio as gr
 
-# --- CONFIGURAZIONE ---
-companies = [
-    "Microsoft", "Apple", "Google", "Amazon", "Facebook",
-    "Tesla", "IBM", "Intel", "Netflix", "Adobe"
-]
+# --- CONFIG ---
+companies = ["Microsoft", "Apple", "Google", "Amazon", "Facebook",
+             "Tesla", "IBM", "Intel", "Netflix", "Adobe"]
 
 TICKERS = {
     "Microsoft": "MSFT",
@@ -26,102 +24,106 @@ TICKERS = {
 
 HF_DATASET = "SelmaNajih001/FT_MultiCompany"
 MODEL_SENTIMENT = "SelmaNajih001/SentimentBasedOnPriceVariation"
-MODEL_PRICE = "SelmaNajih001/PricePredictionForMultiCompany"
+MODEL_PRICE_TESLA = "SelmaNajih001/PricePredictionForTesla"
+MODEL_PRICE_MICROSOFT = "SelmaNajih001/PricePredictionForMicrosoft"
 
 # --- PIPELINES ---
 sentiment_pipeline = pipeline("sentiment-analysis", model=MODEL_SENTIMENT)
-price_pipeline = pipeline("text-classification", model=MODEL_PRICE)
-
-# --- CARICAMENTO DATASET ---
-hf_data = load_dataset(HF_DATASET)["train"]
-df = pd.DataFrame(hf_data)
-
-# --- PREPROCESSING DATI ---
-df['date'] = pd.to_datetime(df['Data'], errors='coerce')
-df['date_merge'] = df['date'].dt.normalize()
-df.sort_values('date', inplace=True)
-
-# --- CALCOLO SENTIMENT ---
-df['Sentiment'] = df['Confidence'] = 0.0
-for i, row in df.iterrows():
-    try:
-        result = sentiment_pipeline(row['Riassunto'])[0]
-        df.at[i, 'Sentiment'] = result['label'].upper().strip()
-        df.at[i, 'Confidence'] = result['score']
-    except Exception:
-        df.at[i, 'Sentiment'] = 'ERROR'
-        df.at[i, 'Confidence'] = 0.0
-
-# --- CALCOLO PREVISIONI PREZZI ---
-df['Predicted'] = 0.0
-for i, row in df.iterrows():
+price_pipeline_tesla = pipeline("text-classification", model=MODEL_PRICE_TESLA)
+price_pipeline_msft = pipeline("text-classification", model=MODEL_PRICE_MICROSOFT)
+
+# --- LOAD DATASET ---
+df_multi = pd.DataFrame(load_dataset(HF_DATASET)["train"])
+df_multi['date'] = pd.to_datetime(df_multi['Data'], errors='coerce')
+df_multi['date_merge'] = df_multi['date'].dt.normalize()
+df_multi.sort_values('date', inplace=True)
+
+# --- SENTIMENT & PREDICTION ---
+df_multi['Sentiment'] = ""
+df_multi['Confidence'] = 0.0
+df_multi['Predicted'] = 0.0
+
+for i, row in df_multi.iterrows():
+    company = row['Company']
+    
+    # Sentiment for all companies
     try:
-        val = price_pipeline(row['Riassunto'])[0]['score']
-        df.at[i, 'Predicted'] = min(val, 1.0)
-    except Exception:
-        df.at[i, 'Predicted'] = 0.0
-
-# --- CARICAMENTO PREZZI AZIONARI ---
-start_date = df['date'].min()
-end_date = pd.Timestamp.today()
-df_prices = pd.DataFrame()
-
+        res = sentiment_pipeline(row['Riassunto'])[0]
+        df_multi.at[i,'Sentiment'] = res['label'].upper().strip()
+        df_multi.at[i,'Confidence'] = res['score']
+    except:
+        df_multi.at[i,'Sentiment'] = 'ERROR'
+        df_multi.at[i,'Confidence'] = 0.0
+    
+    # Regression only for Tesla & Microsoft
+    if company == "Tesla":
+        try:
+            val = price_pipeline_tesla(row['Riassunto'])[0]['score']
+            df_multi.at[i,'Predicted'] = min(val, 1.0)
+        except:
+            df_multi.at[i,'Predicted'] = 0.0
+    elif company == "Microsoft":
+        try:
+            val = price_pipeline_msft(row['Riassunto'])[0]['score']
+            df_multi.at[i,'Predicted'] = min(val, 1.0)
+        except:
+            df_multi.at[i,'Predicted'] = 0.0
+
+# --- FETCH STOCK PRICES ---
+prices = {}
 for company, ticker in TICKERS.items():
-    company_prices = yf.download(ticker, start=start_date, end=end_date)[['Close']].reset_index()
-    company_prices['date_merge'] = pd.to_datetime(company_prices['Date']).dt.normalize()
-    df_prices = pd.merge(df_prices, company_prices, on='date_merge', how='outer') if not df_prices.empty else company_prices
-
-# --- MERGE DATI ---
-df = pd.merge(df, df_prices, on='date_merge', how='inner')
-
-# --- CALCOLO STRATEGIE ---
-df['ValueDaily'] = 0
-df['ValueCumulative'] = 0
-df['ValueDaily_Pred'] = 0
-df['ValueCumulative_Pred'] = 0
-
-for i in range(1, len(df)):
-    score = df.loc[i, 'Confidence']
-    pct = df.loc[i, 'PctChangeDaily'] if pd.notnull(df.loc[i, 'PctChangeDaily']) else 0
-    if df.loc[i, 'Sentiment'] == "UP" and score > 0.8:
-        df.loc[i, 'ValueDaily'] = df.loc[i, 'Close'] * pct
-    elif df.loc[i, 'Sentiment'] == "DOWN" and score > 0.8:
-        df.loc[i, 'ValueDaily'] = df.loc[i, 'Close'] * pct * -1
-    else:
-        df.loc[i, 'ValueDaily'] = 0
-    df.loc[i, 'ValueCumulative'] = df.loc[i-1, 'ValueCumulative'] + df.loc[i, 'ValueDaily']
-
-    df.loc[i, 'ValueDaily_Pred'] = df.loc[i, 'Predicted'] * df.loc[i, 'PctChangeDaily'] * df.loc[i, 'Close']
-    df.loc[i, 'ValueCumulative_Pred'] = df.loc[i-1, 'ValueCumulative_Pred'] + df.loc[i, 'ValueDaily_Pred']
-
-# --- VISUALIZZAZIONE STRATEGIE ---
-df_plot = pd.concat([
-    df[['Data', 'ValueCumulative']].assign(Company='Sentiment'),
-    df[['Data', 'ValueCumulative_Pred']].assign(Company='Regression')
-], ignore_index=True)
-
-fig = px.line(
-    df_plot,
-    x='Data',
-    y='value',
-    color='Company',
-    title="Strategie Sentiment vs Regressione",
-    labels={'value': 'Valore Cumulativo', 'Data': 'Data'}
-)
-
-# --- INTERFACCIA GRADIO ---
-def display_data():
-    return df.head(10), fig
+    start_date = df_multi[df_multi['Company']==company]['date'].min()
+    end_date = pd.Timestamp.today()
+    df_prices = yf.download(ticker, start=start_date, end=end_date)[['Close']].reset_index()
+    df_prices['date_merge'] = pd.to_datetime(df_prices['Date']).dt.normalize()
+    df_prices['PctChangeDaily'] = df_prices['Close'].pct_change().shift(-1)
+    prices[company] = df_prices
+
+# --- MERGE & CALCULATE STRATEGIES ---
+dfs_final = {}
+for company in companies:
+    df_c = df_multi[df_multi['Company']==company].copy()
+    df_c = pd.merge(df_c, prices[company], on='date_merge', how='inner')
+    
+    # Strategy A: Sentiment
+    df_c['StrategyA_Daily'] = 0
+    df_c['StrategyA_Cumulative'] = 0
+    for i in range(1, len(df_c)):
+        score = df_c.loc[i,'Confidence']
+        pct = df_c.loc[i,'PctChangeDaily'] if pd.notnull(df_c.loc[i,'PctChangeDaily']) else 0
+        if df_c.loc[i,'Sentiment']=="UP" and score>0.8:
+            df_c.loc[i,'StrategyA_Daily'] = df_c.loc[i,'Close']*pct
+        elif df_c.loc[i,'Sentiment']=="DOWN" and score>0.8:
+            df_c.loc[i,'StrategyA_Daily'] = df_c.loc[i,'Close']*pct*-1
+        df_c.loc[i,'StrategyA_Cumulative'] = df_c.loc[i-1,'StrategyA_Cumulative'] + df_c.loc[i,'StrategyA_Daily']
+    
+    # Strategy B: Regression (only Tesla & Microsoft)
+    df_c['StrategyB_Daily'] = df_c['Predicted']*df_c['PctChangeDaily']*df_c['Close']
+    df_c['StrategyB_Cumulative'] = df_c['StrategyB_Daily'].cumsum()
+    
+    dfs_final[company] = df_c
+
+# --- PLOT STRATEGIES ---
+df_plot_A = pd.concat([df[['Data','StrategyA_Cumulative']].assign(Company=company) for company, df in dfs_final.items()], ignore_index=True)
+df_plot_B = pd.concat([df[['Data','StrategyB_Cumulative']].assign(Company=company) for company, df in dfs_final.items()], ignore_index=True)
+
+fig_strategy_A = px.line(df_plot_A, x='Data', y='StrategyA_Cumulative', color='Company',
+                         title="Portfolio Evolution Following Strategy A (Sentiment)",
+                         labels={'StrategyA_Cumulative':'Cumulative Value','Data':'Date'})
+fig_strategy_B = px.line(df_plot_B, x='Data', y='StrategyB_Cumulative', color='Company',
+                         title="Portfolio Evolution Following Strategy B (Regression)",
+                         labels={'StrategyB_Cumulative':'Cumulative Value','Data':'Date'})
+
+# --- GRADIO INTERFACE ---
+outputs = [gr.Dataframe(label=company) for company in dfs_final.keys()]
+outputs += [gr.Plot(label="Strategy A: Sentiment"), gr.Plot(label="Strategy B: Regression")]
 
 demo = gr.Interface(
-    fn=display_data,
+    fn=lambda: [df.head(10) for df in dfs_final.values()] + [fig_strategy_A, fig_strategy_B],
     inputs=[],
-    outputs=[
-        gr.Dataframe(label="Dati Aziendali"),
-        gr.Plot(label="Strategia Sentiment vs Regressione")
-    ],
-    title="Analisi Sentiment e Previsioni Prezzi Azionari",
-    description="Visualizza le strategie basate su sentiment e modelli di previsione dei prezzi per diverse aziende."
+    outputs=outputs,
+    title="Portfolio Evolution: Strategy A vs Strategy B",
+    description="Visualizes portfolio evolution based on sentiment (Strategy A) and regression predictions (Strategy B). Regression applies only to Tesla and Microsoft."
 )
 
 demo.launch()