full app

app.py

@@ -0,0 +1,153 @@
+import os
+import numpy as np
+import pandas as pd
+import datetime
+import yfinance as yf
+import streamlit as st
+from sklearn.preprocessing import MinMaxScaler
+from tensorflow.keras.models import load_model
+import matplotlib.pyplot as plt
+
+# ── Disable GPU so TensorFlow runs on CPU ────────────────────────────────
+os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
+
+# ── Mapping sectors to .h5 model filenames ───────────────────────────────
+MODEL_DIR = "models/"
+sector_model_map = {
+    "Technology": "Technology.h5",
+    "Financial Services": "finance_model.h5",
+    "Healthcare": "Healthcare.h5",
+    "Energy": "Energy.h5",
+    "Consumer Defensive": "Consumer Goods.h5",
+    "Consumer Cyclical": "Consumer Services.h5",
+    "Industrials": "industrials.h5",
+    "Real Estate": "Real Estate.h5",
+    "Communication Services": "communication.h5",
+    "Utilities": "Utilities.h5",
+    "Basic Materials": "Material.h5"
+}
+
+# ── Get sector via Yahoo Finance ─────────────────────────────────────────
+def get_sector(ticker: str) -> str:
+    try:
+        info = yf.Ticker(ticker).info
+        return info.get("sector", "Technology")
+    except Exception:
+        return "Technology"
+
+# ── Load Keras .h5 model on CPU without compiling ────────────────────────
+def load_model_keras(sector: str):
+    model_file = sector_model_map.get(sector)
+    if not model_file:
+        return None
+    path = os.path.join(MODEL_DIR, model_file)
+    if not os.path.exists(path):
+        return None
+    return load_model(path, compile=False)
+
+# ── Fetch historical stock data ──────────────────────────────────────────
+def fetch_stock_data(ticker: str, days: int = 50) -> pd.DataFrame | None:
+    end_date = datetime.datetime.today()
+    start_date = end_date - datetime.timedelta(days=days * 2)
+    df = yf.download(ticker, start=start_date, end=end_date)
+    return df.tail(days).reset_index() if not df.empty else None
+
+# ── Scale features and pad if needed ─────────────────────────────────────
+def prepare_data(df: pd.DataFrame):
+    feature_cols = ["Open", "High", "Low", "Close", "Volume"]
+    data = df[feature_cols].values
+    scaler = MinMaxScaler()
+    data_scaled = scaler.fit_transform(data)
+
+    # Pad to 12 features
+    expected_features = 12
+    padded_data = np.pad(
+        data_scaled,
+        ((0, 0), (0, expected_features - data_scaled.shape[1])),
+        mode='constant'
+    )
+
+    return padded_data, scaler
+
+# ── Predict next N days ──────────────────────────────────────────────────
+def predict_next_days(model, data_scaled, seq_length: int, scaler, days: int = 5):
+    preds = []
+    last_seq = np.expand_dims(data_scaled[-seq_length:], axis=0)
+    last_known_row = data_scaled[-1].copy()
+    expected_features = data_scaled.shape[1]
+
+    for _ in range(days):
+        scaled_pred = model.predict(last_seq, verbose=0)[0, 0]
+        new_row = last_known_row.copy()
+        if expected_features >= 4:
+            new_row[3] = scaled_pred  # 'Close'
+
+        inv_input = new_row[:5]
+        inv_scaled = scaler.inverse_transform([inv_input])[0]
+        actual_close = inv_scaled[3]
+        preds.append(actual_close)
+
+        padded = np.pad(inv_scaled, (0, expected_features-5), mode='constant')
+        last_seq = np.concatenate([last_seq[:,1:,:], padded.reshape(1,1,-1)], axis=1)
+        last_known_row = padded
+
+    return preds
+
+# ── Streamlit UI ─────────────────────────────────────────────────────────
+st.set_page_config(page_title="📈 Stock Predictor", layout="centered")
+st.title("📊 Sector‑Based Stock Price Predictor ")
+
+ticker = st.text_input("Enter Stock Ticker (e.g. AAPL, MSFT, TSLA):").upper()
+if st.button("Predict"):
+    if not ticker:
+        st.warning("⚠️ Please enter a stock ticker.")
+        st.stop()
+
+    df = fetch_stock_data(ticker)
+    if df is None or len(df) < 50:
+        st.error("❌ Not enough data to predict. Try another ticker.")
+        st.stop()
+
+    sector = get_sector(ticker)
+    model = load_model_keras(sector)
+    if model is None:
+        st.error(f"❌ No model found for sector: {sector}")
+        st.stop()
+    st.success(f"✅ Sector: {sector}")
+
+    data_scaled, scaler = prepare_data(df)
+    preds = predict_next_days(model, data_scaled, seq_length=50, scaler=scaler, days=5)
+
+    # Prepare dates
+    df["Date"] = pd.to_datetime(df["Date"])
+    actual_dates = df["Date"].tail(30)
+    actual_prices = df["Close"].tail(30).values
+
+    last_date = actual_dates.iloc[-1]
+    future_dates = pd.date_range(start=last_date + pd.Timedelta(days=1), periods=5, freq="B")
+
+    # Show result table
+    result_df = pd.DataFrame({
+        "Date": future_dates,
+        "Predicted Close Price": np.round(preds, 2)
+    })
+    st.write(result_df)
+
+    # Plot with Matplotlib
+    fig, ax = plt.subplots(figsize=(10, 5))
+    # Past 30 days
+    ax.plot(actual_dates, actual_prices,
+            color="blue", marker="o", linestyle="-", label="Actual (30d)", markersize=8)
+    # Next 5 days
+    ax.plot(future_dates, preds,
+            color="red", marker="o", linestyle="-", label="Predicted (5d)", markersize=8)
+    # Formatting
+    ax.set_title(f"{ticker} Close Price: Last 30 Days & Next 5 Days Forecast")
+    ax.set_xlabel("Date")
+    ax.set_ylabel("Close Price")
+    ax.legend()
+    fig.autofmt_xdate()
+    ax.grid(True)
+
+    st.pyplot(fig)
+

models/Consumer Goods.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9514605eda1e29a8ce7661d51ff7679fc58b4a91b705baec42bb8876c57de09c
+size 1611744

models/Consumer Services.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1278246ded80173c42fce1eafcb3680c30324abd744887f9cd5af011136649a
+size 1611808

models/Energy.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:644deb96cbb912d987f7885572ea58d47757f3d232e391e53c0e8d0d7f699905
+size 1611744

models/Healthcare.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e70abe03522cec27eb4eef00caadfe44950cfae07d78ac748a61dc0b317e72e
+size 1611744

models/Material.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4eeea5515544bfdac2b73a815711d3889fda8282269c3554c02f46c33f1a4cf
+size 1611808

models/Real Estate.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25606d9b7a9d67e8a389294e63a1254f3e32a0fa84cced94ffdfb0520cd7e4a8
+size 1611808

models/Technology.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f708e973bd025439c61786849c190a5a7663e7b4efe80e682471e858b9b95f6
+size 1613640

models/Utilities.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57babfe3b7edbc4afdd44c3eae25ba4e896ae40c363a2e3eebc49b3813f8eabb
+size 1611808

models/communication.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1851bf2ad94764cf60a53bea705b36bf064791fc127e11d360620f5306d697b
+size 1611808

models/finance_model.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26937b0f6b8eef46e2f9737c78dfa669c0c3b0e324f054b2a061bdd4523728c1
+size 1611744

models/industrials.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fcba5c9f11d2a96a5d8bf7c4f1816b2a4ad8d9758b0ad9c8f20f01bbd3db322
+size 1611808