init

Simulation.py

@@ -6,6 +6,7 @@ import os
 
 from hts.HTS import HTS
 from stock.util.Stock2Vector import Stock2Vector
+from stock.util.StockPredictor import StockPredictor
 from hts.BuySellChecker import BuySellChecker
 
 class Simulation (HTS):
@@ -16,6 +17,7 @@ class Simulation (HTS):
         super().__init__(RESOURCE_PATH)
 
         self.stock2Vector = Stock2Vector(RESOURCE_PATH)
+        self.stockPredictor = StockPredictor()
         self.buySellChecker = BuySellChecker()
         self.RESOURCE_PATH = RESOURCE_PATH
         #self.connect()
@@ -126,23 +128,23 @@ class Simulation (HTS):
 
         return
 
-    def analyzeAutoMode(self, data):
-        return data, None
 
-    def simulate(self, stock_code, today, type="rule"):
+    def simulate(self, stock_code, today, method="rule"):
-        LAST_DATA = self.stock2Vector.getLastData(stock_code, today)
 
-        result = self.stock2Vector.getRealTime(stock_code, today, LAST_DATA)
+        if method == "ml":
+            LAST_DATA = self.stock2Vector.getLastData(stock_code, today, n=10)
+            result = self.stock2Vector.getRealTime(stock_code, today, LAST_DATA)
+
+            df, minmax_df = self.stock2Vector.preprocessData(result)
+            bsLine, data = self.stockPredictor.predict(df, minmax_df, isRealTime=False)
+        else:
+            LAST_DATA = self.stock2Vector.getLastData(stock_code, today)
+            result = self.stock2Vector.getRealTime(stock_code, today, LAST_DATA)
 
-        if type == "rule":
             # 이동평균, RSI, MACD, 일목균형, 볼린저밴드 상/하단을 계산한다.
             data = self.buySellChecker.analyze(result)
             # 사야 할 시점과 팔아야 할 시점을 체크한다.
             bsLine, data = self.buySellChecker.checkTransaction(data, stock_code, isRealTime=False)
-        elif type == "auto":
-            data, bsLine = self.analyzeAutoMode(result)
-        else:
-            data, bsLine = None, None
 
         if data is not None:
             # 그래프를 그린다.
@@ -159,13 +161,13 @@ if __name__ == "__main__":
     stock_codes = {
         # 252670
         # 122630
-        "122630": ['20220729'],
+        "122630": ['20220803'],
     }
 
     for stock_code in stock_codes:
         simulation = Simulation(RESOURCE_PATH)
 
         for given_day in stock_codes[stock_code]:
-            simulation.simulate(stock_code, given_day)
+            simulation.simulate(stock_code, given_day, method='ml')
 
     print ("done...")

StockTrainer.py

@@ -1,7 +1,6 @@
 import os
 import keras
 import numpy as np
-from numpy import zeros, newaxis
 import tensorflow as tf
 from stock.util.Stock2Vector import Stock2Vector
 from classification_models.keras import Classifiers
@@ -18,16 +17,17 @@ class StockTrainer:
 
     def getDataset(self, stock_code):
         VECTOR_SIZE = 299
-        df, minmax_df = self.stock2Vector.makeTrainData(stock_code)
+        result = self.stock2Vector.getTrainData(stock_code)
+        df, minmax_df = self.stock2Vector.preprocessData(result)
 
         TOTAL_X, TOTAL_Y = [], []
-        for key in df:
+        for key in minmax_df:
             if key == "date":
                 continue
             elif key == "label":
-                TOTAL_Y.append(df[key].tolist())
+                TOTAL_Y.append(minmax_df[key].tolist())
             else:
-                TOTAL_X.append(df[key].tolist())
+                TOTAL_X.append(minmax_df[key].tolist())
 
         SIZE_WIDTH = len(TOTAL_X[0])
         SIZE_HEIGHT = len(TOTAL_X)
@@ -44,38 +44,37 @@ class StockTrainer:
                 Y.append([0, 1, 0])
             else:
                 Y.append([0, 0, 1])
-            if i >= VECTOR_SIZE+10:
-                break
 
         X = np.asarray(X)
         Y = np.asarray(Y)
         return X, Y
 
     def train(self, stock_code):
-        ResNet18, preprocess_input = Classifiers.get('inceptionresnetv2')
-
         X, Y = self.getDataset(stock_code)
 
+        # build model
+        n_classes = 3
+        Inceptionresnetv2, preprocess_input = Classifiers.get('inceptionresnetv2')
         X = preprocess_input(X)
 
-        n_classes = 3
+        # train
-
+        checkpoint_filename = os.path.join(self.RESOURCE_PATH, "model", "stock.ckpt")
-        # build model
+        base_model = Inceptionresnetv2(input_shape=(299, 299, 3), include_top=False)
-        base_model = ResNet18(input_shape=(299, 299, 3), include_top=False)
         x = tf.keras.layers.GlobalAveragePooling2D()(base_model.output)
         output = tf.keras.layers.Dense(n_classes, activation='softmax')(x)
         model = keras.models.Model(inputs=[base_model.input], outputs=[output])
 
-        # train
         model.compile(optimizer='SGD', loss='categorical_crossentropy', metrics=['accuracy'])
-
-        checkpoint_filename = os.path.join(self.RESOURCE_PATH, "model", "stock.ckpt")
         chekpoint = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_filename, save_weights_only=True, verbose=1)
-        earlystop = tf.keras.callbacks.EarlyStopping(monitor='loss', patience=3, mode="auto")
+        earlystop = tf.keras.callbacks.EarlyStopping(monitor='loss', patience=5, mode="auto")
+
+        if os.path.isfile(checkpoint_filename):
+            model.load_weights(checkpoint_filename)
 
         model.fit(x=X,
                   y=Y,
-                  epochs=10,
+                  batch_size=10000,
+                  epochs=10000,
                   callbacks=[chekpoint, earlystop])
 
         return
@@ -95,5 +94,4 @@ if __name__ == "__main__":
         stockTrainer = StockTrainer(RESOURCE_PATH)
         stockTrainer.train(stock_code)
 
-
     print ("done...")

hts/HTS.py

@@ -634,7 +634,7 @@ class HTS:
                   "label": []}
 
         days = []
-        for i in range(1, 10):
+        for i in range(1, 100):
             last_day = (datetime.strptime(today, '%Y%m%d') - timedelta(i)).strftime('%Y%m%d')
             isValid = self.isValidYMD(stock_code, last_day)
             if isValid:
@@ -655,6 +655,7 @@ class HTS:
         else:
             result = {"check": set(), "time": [], "open": [], "close": [], "high": [], "low": [], "vol": [], "label": []}
 
+        #### real time에서 아직 저장된 것이 없기 때문에 result는 아무것도 채워지지 않는다.
         self.getDBData(stock_code, today, result)
 
 

stock/util/Stock2Vector.py

@@ -153,6 +153,53 @@ class Stock2Vector(HTS):
 
         return df, minmax_df
 
+    def getTrainData(self, stock_code):
+        tableName = 'hts'
+        conn = sqlite3.connect(os.path.join(self.RESOURCE_PATH, "hts.db"))
+        cursor = conn.cursor()
+
+        cursor.execute(
+            'SELECT ymd, hms, open, high, low, close, volume, label FROM ' + tableName + ' WHERE CODE=? and (ymd >= ? and  ymd <= ?) order by ymd desc, hms ',
+            (stock_code, "20220721", "20220731"))
+        db_result = cursor.fetchall()
+        temp_result = []
+        for rows in db_result:
+            temp_result.append(
+                [rows[0], rows[1], rows[2], rows[3], rows[4], rows[5], rows[6], 0 if rows[7] is None else rows[7]])
+        temp_result.sort(key=lambda x: (x[0], x[1]))
+
+        result = {"check": set(), "time": [], "open": [], "close": [], "high": [], "low": [], "vol": [], "label": []}
+        for rows in temp_result:
+            ymd = rows[0]  # hts.날짜
+            hms = rows[1]  # hts.시간
+            open = rows[2]  # hts.시가
+            high = rows[3]  # hts.고가
+            low = rows[4]  # hts.저가
+            close = rows[5]  # hts.종가
+            vol = rows[6]  # hts.거래량
+            label = 0 if rows[7] is None else rows[7]  # hts.매매구분
+
+            temp = datetime.strptime(str(ymd) + " " + str(hms).zfill(4) + "00", '%Y%m%d %H%M%S')
+
+            result["time"].append(temp)
+            result["open"].append(int(open))
+            result["close"].append(int(close))
+            result["high"].append(int(high))
+            result["low"].append(int(low))
+            result["vol"].append(int(vol))
+            result["label"].append(int(label))
+
+        return result
+    def preprocessData(self, result):
+        # 분석을 통해서 볼린저밴드 상/하단을 계산한다.
+        df = self.buySellChecker.getVectorFeature(result)
+        minmax_df1 = (df - df.min()) / (df.max() - df.min())
+        minmax_df2 = minmax_df1.drop(["date"], axis="columns")
+        minmax_df = minmax_df2.join(df['date'])
+
+        minmax_df = minmax_df.fillna(0)
+        return df, minmax_df
+
     def makeTrainData(self, stock_code):
         result = {"check": set(), "time": [], "open": [], "close": [], "high": [], "low": [], "vol": [], "label": []}
 

stock/util/StockPredictor.py

@@ -0,0 +1,59 @@
+import os
+import keras
+import numpy as np
+from keras.applications.imagenet_utils import decode_predictions
+from classification_models.keras import Classifiers
+
+class StockPredictor:
+
+    RESOURCE_PATH = None
+    stock2Vector = None
+
+    def __init__(self):
+        return
+
+    def getDataset(self, df):
+        VECTOR_SIZE = 299
+        TOTAL_X, TOTAL_Y = [], []
+        for key in df:
+            if key == "date":
+                continue
+            elif key == "label":
+                TOTAL_Y.append(df[key].tolist())
+            else:
+                TOTAL_X.append(df[key].tolist())
+
+        SIZE_WIDTH = len(TOTAL_X[0])
+        SIZE_HEIGHT = len(TOTAL_X)
+        X = []
+        for i in range(VECTOR_SIZE, SIZE_WIDTH):
+            temp_X, temp_Y = np.zeros((VECTOR_SIZE, VECTOR_SIZE)), np.zeros(0)
+            for j in range(SIZE_HEIGHT):
+                temp_X[j][0:VECTOR_SIZE] = TOTAL_X[j][i - VECTOR_SIZE:i]
+            temp_X = np.stack([temp_X, temp_X, temp_X], axis=-1)
+            X.append(temp_X)
+
+        X = np.asarray(X[len(X)-1])
+
+        return X
+
+    def predict(self, df, minmax_df, isRealTime=False):
+        X = self.getDataset(df)
+
+        # build model
+        n_classes = 3
+        Inceptionresnetv2, preprocess_input = Classifiers.get('inceptionresnetv2')
+        X = preprocess_input(X)
+        base_model = Inceptionresnetv2(input_shape=(299, 299, 3), include_top=False)
+        model = keras.models.Model(inputs=[base_model.input])
+
+        checkpoint_filename = os.path.join(self.RESOURCE_PATH, "model", "stock.ckpt")
+        model.load_weights(checkpoint_filename)
+
+        y = model.predict(X)
+
+        # result
+        print(decode_predictions(y))
+
+
+        return