Update app.py

app.py

@@ -1,474 +1,240 @@
-
 import os
-import time
-import tempfile
+import io
 import requests
+import numpy as np
 import pandas as pd
 import gradio as gr
-
-from datetime import datetime, timedelta, timezone
+import xgboost as xgb
+from sklearn.neighbors import KNeighborsClassifier
+from sklearn.preprocessing import StandardScaler
 from groq import Groq
 
 # ==========================================
-# CONFIGURATION
+# 🔌 1. GROQ CLIENT
 # ==========================================
-
-OKX_URL = "https://www.okx.com/api/v5/market/history-candles"
-
-GROQ_API_KEY = os.getenv("GROQ_API_KEY")
-
-GROQ_MODEL = os.getenv(
-    "GROQ_MODEL",
-    "openai/gpt-oss-120b"
-)
+api_key = os.environ.get("GROQ_API_KEY")
+client = None
+if api_key:
+    client = Groq(api_key=api_key)
 
 # ==========================================
-# OKX PUBLIC DATA
+# 📊 2. YARDIMCI FONKSİYONLAR
 # ==========================================
-
-def get_okx_data(days, timeframe, symbol, progress=gr.Progress()):
-
-    days = int(days)
-    symbol = symbol.strip().upper()
-
-    if not symbol:
-        raise ValueError("Trading pair gerekli.")
-
-    end = datetime.now(timezone.utc)
-    start = end - timedelta(days=days)
-
-    start_ms = int(start.timestamp() * 1000)
-    end_ms = int(end.timestamp() * 1000)
-
-    all_data = []
-    seen = set()
-    after = None
-
-    session = requests.Session()
-
-    while True:
-
-        params = {
-            "instId": symbol,
-            "bar": timeframe,
-            "limit": "300"
-        }
-
-        if after is not None:
-            params["after"] = str(after)
-
-        response = session.get(
-            OKX_URL,
-            params=params,
-            timeout=30
-        )
-
-        response.raise_for_status()
-
-        result = response.json()
-
-        if result.get("code") != "0":
-            raise RuntimeError(
-                result.get("msg", "OKX API error")
-            )
-
-        data = result.get("data", [])
-
-        if not data:
-            break
-
-        for candle in data:
-
-            ts = int(candle[0])
-
-            if start_ms <= ts <= end_ms and ts not in seen:
-                seen.add(ts)
-                all_data.append(candle)
-
-        oldest = min(int(x[0]) for x in data)
-
-        progress(
-            min(
-                0.99,
-                max(
-                    0.01,
-                    1 - (oldest - start_ms) /
-                    max(1, end_ms - start_ms)
-                )
-            ),
-            desc=f"OKX candles: {len(all_data)}"
-        )
-
-        if oldest <= start_ms:
-            break
-
-        after = oldest
-
-        time.sleep(0.15)
-
-    if not all_data:
-        raise ValueError("OKX verisi bulunamadı.")
-
-    columns = [
-        "timestamp",
-        "open",
-        "high",
-        "low",
-        "close",
-        "volume",
-        "volume_currency",
-        "volume_quote",
-        "confirm"
-    ]
-
-    df = pd.DataFrame(
-        all_data,
-        columns=columns
-    )
-
-    df["timestamp"] = pd.to_datetime(
-        df["timestamp"].astype("int64"),
-        unit="ms",
-        utc=True
-    )
-
-    numeric_columns = [
-        "open",
-        "high",
-        "low",
-        "close",
-        "volume",
-        "volume_currency",
-        "volume_quote"
-    ]
-
-    for col in numeric_columns:
-        df[col] = pd.to_numeric(df[col])
-
-    df["confirm"] = pd.to_numeric(df["confirm"])
-
-    df = (
-        df.sort_values("timestamp")
-        .drop_duplicates("timestamp")
-        .reset_index(drop=True)
+def compute_rsi(series, period=14):
+    delta = series.diff()
+    gain = delta.clip(lower=0).rolling(period).mean()
+    loss = -delta.clip(upper=0).rolling(period).mean()
+    rs = gain / loss
+    return 100 - (100 / (1 + rs))
+
+def load_csv_from_url(url):
+    try:
+        headers = {"User-Agent": "Mozilla/5.0"}
+        r = requests.get(url, headers=headers, timeout=20)
+        r.raise_for_status()
+        df = pd.read_csv(io.StringIO(r.text))
+        return df, None
+    except Exception as e:
+        return None, f"CSV indirilemedi: {e}"
+
+def process_dataframe(df):
+    df.columns = [c.strip().lower() for c in df.columns]
+    col_map = {
+        "date": "date", "datetime": "date", "timestamp": "date",
+        "open": "open", "high": "high", "low": "low",
+        "close": "close", "adj close": "close", "price": "close",
+        "volume": "volume", "vol": "volume"
+    }
+    df = df.rename(columns={c: col_map[c] for c in df.columns if c in col_map})
+
+    if "close" not in df.columns:
+        return None, None, None, f"'close' sütunu bulunamadı. Mevcut sütunlar: {list(df.columns)}"
+
+    for col in ["open", "high", "low", "close", "volume"]:
+        if col in df.columns:
+            df[col] = pd.to_numeric(df[col], errors="coerce")
+
+    df = df.dropna(subset=["close"]).reset_index(drop=True)
+    if len(df) < 50:
+        return None, None, None, "CSV'de en az 50 satır veri olmalı."
+
+    # Teknik göstergeler
+    df["ma_5"] = df["close"].rolling(5).mean()
+    df["ma_20"] = df["close"].rolling(20).mean()
+    df["rsi"] = compute_rsi(df["close"], 14)
+    df["macd"] = df["close"].ewm(span=12).mean() - df["close"].ewm(span=26).mean()
+    df["target"] = (df["close"].shift(-1) > df["close"]).astype(int)
+
+    df_model = df.dropna().reset_index(drop=True)
+    if len(df_model) < 20:
+        return None, None, None, "Göstergeler hesaplandıktan sonra yeterli veri kalmadı."
+
+    features = ["ma_5", "ma_20", "rsi", "macd"]
+    if "volume" in df_model.columns:
+        features.append("volume")
+
+    X = df_model[features].values
+    y = df_model["target"].values
+
+    # --- XGBoost ---
+    xgb_model = xgb.XGBClassifier(
+        n_estimators=50, max_depth=3, learning_rate=0.1,
+        objective="binary:logistic", eval_metric="logloss", random_state=42
     )
-
-    return df
-
+    xgb_model.fit(X, y)
+
+    # --- KNN (ölçeklendirme gerekli) ---
+    scaler = StandardScaler()
+    X_scaled = scaler.fit_transform(X)
+    knn_model = KNeighborsClassifier(n_neighbors=5)
+    knn_model.fit(X_scaled, y)
+
+    latest = df_model.iloc[-1]
+    metrics = {
+        "last_close": float(latest["close"]),
+        "ma5": float(latest["ma_5"]),
+        "ma20": float(latest["ma_20"]),
+        "rsi": float(latest["rsi"]),
+        "macd": float(latest["macd"]),
+        "features": features,
+    }
+    if "volume" in df_model.columns:
+        metrics["volume"] = float(latest["volume"])
+
+    return (xgb_model, knn_model, scaler), metrics, features, None
 
 # ==========================================
-# TECHNICAL INDICATORS
+# 🧠 3. ANALİZ MOTORU (Ensemble: XGB + KNN)
 # ==========================================
+def predict_and_analyze(csv_url):
+    if not csv_url:
+        return "Lütfen bir CSV linki giriniz.", {}
 
-def add_indicators(df):
-
-    df = df.copy()
+    df, err = load_csv_from_url(csv_url)
+    if err:
+        return err, {}
 
-    df["return_pct"] = (
-        df["close"].pct_change() * 100
-    )
-
-    df["EMA20"] = df["close"].ewm(
-        span=20,
-        adjust=False
-    ).mean()
-
-    df["EMA50"] = df["close"].ewm(
-        span=50,
-        adjust=False
-    ).mean()
+    models, metrics, features, err = process_dataframe(df)
+    if err:
+        return err, {}
 
-    delta = df["close"].diff()
+    xgb_model, knn_model, scaler = models
 
-    gain = delta.clip(lower=0).ewm(
-        alpha=1/14,
-        adjust=False
-    ).mean()
+    feat_values = [metrics[k] for k in ["ma5", "ma20", "rsi", "macd"]]
+    if "volume" in metrics:
+        feat_values.append(metrics["volume"])
+    feat_arr = np.array([feat_values])
 
-    loss = (-delta.clip(upper=0)).ewm(
-        alpha=1/14,
-        adjust=False
-    ).mean()
+    # XGBoost tahmini
+    xgb_prob_up = float(xgb_model.predict_proba(feat_arr)[0][1])
 
-    rs = gain / loss.replace(0, float("nan"))
+    # KNN tahmini (ölçeklendirilmiş veriyle)
+    feat_scaled = scaler.transform(feat_arr)
+    knn_prob_up = float(knn_model.predict_proba(feat_scaled)[0][1])
 
-    df["RSI"] = 100 - (100 / (1 + rs))
+    # Ensemble (ortalama)
+    ensemble_prob_up = (xgb_prob_up + knn_prob_up) / 2
+    ensemble_prob_down = 1 - ensemble_prob_up
 
-    previous_close = df["close"].shift(1)
+    direction = "YUKARI 📈" if ensemble_prob_up > 0.5 else "AŞAĞI 📉"
+    confidence = max(ensemble_prob_up, ensemble_prob_down) * 100
 
-    tr = pd.concat(
-        [
-            df["high"] - df["low"],
-            (df["high"] - previous_close).abs(),
-            (df["low"] - previous_close).abs()
-        ],
-        axis=1
-    ).max(axis=1)
-
-    df["ATR"] = tr.ewm(
-        alpha=1/14,
-        adjust=False
-    ).mean()
-
-    df["volume_sma"] = (
-        df["volume"].rolling(20).mean()
-    )
-
-    return df
-
-
-# ==========================================
-# GROQ GPT-OSS AI
-# ==========================================
-
-def groq_analysis(df):
-
-    if not GROQ_API_KEY:
-        raise ValueError(
-            "GROQ_API_KEY bulunamadı. "
-            "Hugging Face Secrets bölümüne ekle."
-        )
-
-    client = Groq(
-        api_key=GROQ_API_KEY
-    )
-
-    latest = df.tail(30).copy()
-
-    market_data = latest[
-        [
-            "timestamp",
-            "open",
-            "high",
-            "low",
-            "close",
-            "volume",
-            "return_pct",
-            "EMA20",
-            "EMA50",
-            "RSI",
-            "ATR"
-        ]
-    ].round(4).to_string(index=False)
+    prob_dict = {
+        "XGBoost Yükseliş": xgb_prob_up,
+        "KNN Yükseliş": knn_prob_up,
+        "Ensemble Yükseliş (Ortalama)": ensemble_prob_up,
+    }
 
     prompt = f"""
-BTC-USDT teknik piyasa analizi yap.
-
-VERİLER:
-{market_data}
-
-ANALİZ GÖREVLERİ:
-
-1. Genel piyasa trendi.
-2. EMA20 ve EMA50 karşılaştırması.
-3. RSI momentum analizi.
-4. ATR volatilite analizi.
-5. Hacim değerlendirmesi.
-6. Destek ve direnç seviyeleri.
-7. Bullish senaryosu.
-8. Bearish senaryosu.
-9. Neutral senaryosu.
-10. Risk analizi.
-
-ÇIKTI FORMATI:
-
-YÖN: BULLISH / BEARISH / NEUTRAL
-
-MODEL GÜVENİ: 0-100
-
-TREND:
-MOMENTUM:
-VOLATİLİTE:
-DESTEK:
-DİRENÇ:
-
-YÜKSELİŞ SENARYOSU:
-DÜŞÜŞ SENARYOSU:
-YATAY SENARYO:
-
-RİSK:
-NİHAİ KARAR:
-
-Kurallar:
-- Sadece verilen verilere dayan.
-- Gelecekteki fiyatları uydurma.
-- Güven puanını kalibre edilmiş olasılık gibi sunma.
-- Kesin kazanç garantisi verme.
-- Türkçe yanıt ver.
-"""
-
-    response = client.chat.completions.create(
-        model=GROQ_MODEL,
-        messages=[
-            {
-                "role": "system",
-                "content": (
-                    "You are a professional quantitative "
-                    "crypto market analyst. "
-                    "Analyze data carefully and respond in Turkish."
-                )
-            },
-            {
-                "role": "user",
-                "content": prompt
-            }
-        ],
-        temperature=0.2,
-        max_completion_tokens=2000
+    Sen uzman bir kripto para analisti ve teknik analiz uzmanısın.
+
+    📊 BTC/USD Son Kapanış (CSV'den): ${metrics['last_close']:,.2f}
+    - RSI (14): {metrics['rsi']:.2f}
+    - MA5: ${metrics['ma5']:,.2f} | MA20: ${metrics['ma20']:,.2f}
+    - MACD: {metrics['macd']:.4f}
+    {f"- Hacim: {metrics['volume']:,.2f}" if 'volume' in metrics else ""}
+
+    🤖 Model Sonuçları:
+    - XGBoost Yükseliş Olasılığı: %{xgb_prob_up*100:.1f}
+    - KNN Yükseliş Olasılığı: %{knn_prob_up*100:.1f}
+    - Ensemble (Ortalama) Yükseliş Olasılığı: %{ensemble_prob_up*100:.1f}
+    - Tahmin Yönü: {direction} (Güven: %{confidence:.1f})
+
+    Türkçe bir teknik analiz raporu hazırla. Şu başlıklar bulunsun:
+    1. 🎯 Model Yorumu (XGBoost ve KNN'in karşılaştırmalı mantığı, ensemble neden kullanılır)
+    2. 📈 Teknik Analiz (RSI, MA, MACD; destek/direnç)
+    3. 💡 İşlem Stratejisi (Long/Short, stop-loss, take-profit, risk yönetimi)
+
+    Yatırım tavsiyesi olmadığını vurgula. Rasyonel ve profesyonel ol.
+    """
+
+    if client:
+        try:
+            response = client.chat.completions.create(
+                model="openai/gpt-oss-120b",
+                messages=[{"role": "user", "content": prompt}],
+                temperature=0.3,
+                max_tokens=1024
+            )
+            analysis_result = response.choices[0].message.content
+        except Exception as e:
+            analysis_result = f"Groq API Hatası: {e}"
+    else:
+        analysis_result = "GROQ_API_KEY bulunamadı! Hugging Face Secrets'a ekleyin."
+
+    header = (
+        f"**Son Kapanış: ${metrics['last_close']:,.2f}** | RSI: {metrics['rsi']:.1f} | "
+        f"MA5: ${metrics['ma5']:,.0f} | MA20: ${metrics['ma20']:,.0f}\n\n"
+        f"**XGB:** %{xgb_prob_up*100:.1f} | **KNN:** %{knn_prob_up*100:.1f} | "
+        f"**Ensemble:** %{ensemble_prob_up*100:.1f}\n\n---\n\n"
     )
-
-    return response.choices[0].message.content
-
-
-# ==========================================
-# MAIN ENGINE
-# ==========================================
-
-def run_system(days, timeframe, symbol):
-
-    try:
-
-        df = get_okx_data(
-            days,
-            timeframe,
-            symbol
-        )
-
-        df = add_indicators(df)
-
-        analysis = groq_analysis(df)
-
-        temp = tempfile.NamedTemporaryFile(
-            suffix=".csv",
-            prefix="OKX_BTC_",
-            delete=False
-        )
-
-        csv_path = temp.name
-        temp.close()
-
-        df.to_csv(
-            csv_path,
-            index=False
-        )
-
-        last = df.iloc[-1]
-
-        info = f"""
-## MARKET INFORMATION
-
-- Symbol: {symbol}
-- Timeframe: {timeframe}
-- Candles: {len(df)}
-- Last Close: {last['close']:.2f}
-- RSI: {last['RSI']:.2f}
-- EMA20: {last['EMA20']:.2f}
-- EMA50: {last['EMA50']:.2f}
-- ATR: {last['ATR']:.2f}
-- Volume: {last['volume']:.4f}
-"""
-
-        return (
-            info,
-            analysis,
-            csv_path,
-            df.tail(20)
-        )
-
-    except Exception as e:
-
-        return (
-            f"ERROR: {str(e)}",
-            "",
-            None,
-            pd.DataFrame()
-        )
-
+    return header + analysis_result, prob_dict
 
 # ==========================================
-# GRADIO INTERFACE
+# 🎨 4. GRADIO ARAYÜZÜ
 # ==========================================
-
-with gr.Blocks(
-    title="OKX GPT-OSS BTC AI"
-) as demo:
-
-    gr.Markdown("""
-    # BTC-USDT AI ANALYSIS ENGINE
-
-    ### OKX Public API + Groq GPT-OSS 120B
-
-    Historical Data | Technical Indicators | AI Analysis | CSV
-    """)
-
-    with gr.Row():
-
-        symbol = gr.Textbox(
-            label="Trading Pair",
-            value="BTC-USDT"
-        )
-
-        timeframe = gr.Dropdown(
-            choices=[
-                "1m",
-                "3m",
-                "5m",
-                "15m",
-                "30m",
-                "1H",
-                "2H",
-                "4H",
-                "6H",
-                "12H",
-                "1D"
-            ],
-            value="1H",
-            label="Timeframe"
-        )
-
-        days = gr.Slider(
-            minimum=1,
-            maximum=90,
-            value=30,
-            step=1,
-            label="Historical Days"
-        )
-
-    run = gr.Button(
-        "RUN GPT-OSS BTC AI",
-        variant="primary"
+with gr.Blocks(theme=gr.themes.Soft(primary_hue="orange", secondary_hue="slate")) as demo:
+    gr.Markdown(
+        """
+        # ₿ BTC/USD XGBoost + KNN & Groq AI Tahmin Radarı (CSV Linki)
+        Bir CSV linki girin; kod dosyayı indirir, **XGBoost** ve **KNN** modellerini eğitir, ensemble tahmini üretir ve **Groq** analiz raporu hazırlar.
+        """
     )
 
-    market_info = gr.Markdown()
+    with gr.Row():
+        with gr.Column():
+            gr.Markdown("### 🔗 CSV Linki")
+            url_input = gr.Textbox(
+                label="CSV URL",
+                placeholder="https://.../btc_usd.csv",
+                value=""
+            )
+            gr.Markdown(
+                """
+                **İpucu**: `yfinance` ile indirdiğiniz CSV'yi GitHub'a yükleyip "Raw" linkini buraya yapıştırın.
+                """
+            )
+            submit_btn = gr.Button("CSV'yi İndir ve Analiz Et", variant="primary")
 
-    ai_output = gr.Markdown()
+        with gr.Column():
+            gr.Markdown("### 📈 Model Tahminleri (XGB + KNN)")
+            prob_output = gr.Label(label="Olasılık Dağılımı")
 
-    csv_output = gr.File(
-        label="Download CSV"
-    )
+            gr.Markdown("### 🧠 Groq Yapay Zeka Analiz Raporu")
+            analysis_output = gr.Markdown(value="CSV linkini girip butona basın.")
 
-    preview = gr.Dataframe(
-        label="Last 20 Candles",
-        interactive=False
+    submit_btn.click(
+        fn=predict_and_analyze,
+        inputs=[url_input],
+        outputs=[analysis_output, prob_output]
     )
 
-    run.click(
-        fn=run_system,
-        inputs=[
-            days,
-            timeframe,
-            symbol
-        ],
-        outputs=[
-            market_info,
-            ai_output,
-            csv_output,
-            preview
-        ]
+    gr.Markdown(
+        """
+        ---
+        ⚠️ **Yatırım Tavsiyesi Değildir**: Bu uygulama eğitim amaçlıdır.
+        """
     )
 
-
-demo.queue().launch(
-    server_name="0.0.0.0",
-    server_port=7860
-)
+if __name__ == "__main__":
+    demo.launch()